(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Systems analysis for business management"



:v:-.:\'.'tuV>ii- L .'.;i:I 



for BUSINESS MANAGEMENT 



UNIVERSITY 
OF FLORIDA 
LIBRARIES 




ENGINEERING AND PHYSICS 



LI BRARY 



Systems Analysis 

for 
Business Management 



^nl 



Stanford L. Optner 

Consultant in Systems Engineering 
and data processing, Stanford L. 
Optner and associates; member 
Of the instructional staff, Engineering 
Extension, University of California 
at Los Angeles 



I960 



Prentice-Hall, Inc. 



Englewood Cliffs, N.J, 



© 1960 
PRENTICE-HALL, INC. 
Englewood Cliffs, N.J. 



ALL RIGHTS RESERVED. NO PART OF THIS BOOK 
MAY BE REPRODUCED IN ANY FORM, BY MIMEO- 
GRAPH OR ANY OTHER MEANS, WITHOUT PER- 
MISSION IN WRITING FROM THE PUBLISHERS. 

LIBRARY OF CONGRESS 
CATALOG CARD NO.: 60-9174 



PRINTED IN THE UNITED STATES OF AMERICA 

88122-C 



To Ruth 



Digitized by the Internet Archive 
in 2013 



http://archive.org/details/systemsanalysisOOoptn 



Preface 



Today, there is no general systems theory which has been applied to 
business. In almost all institutions, business is taught much as it al- 
ways has been. It is among our most undisciplined areas of knowl- 
edge and has exhibited slow, uninspired improvement over the last 
fifty years. Our future business managers are being taught vocational 
skills in the absence of "think processes" which would equip them 
far better for reality problems. A "think process" in this context 
would be a tool of analysis or synthesis which would enable the com- 
petent executive to solve his problems with a high degree of reliabil- 
ity. General systems theory supplies such a vehicle. 

The need for a general systems theory has become increasingly 
evident. Some companies are so large they are no longer understand- 
able in the simple, descriptive terms of their formative years. As a 
result, corporate objectives frequently work at cross purposes with 
current policies. The typical executive is forced to attack his prob- 
lems as if they were an endless number of special cases. What is needed 
are general concepts about the nature of business operations that 
would assist him to abstract the properties of an individual problem. 
A tool of this kind would make the problems of business more acces- 
sible to the analyst. 

Our expanded industrial technological frontiers have not been 
reinforced by the development of large numbers of business trained, 
technically oriented personnel. Industry needs more men who have 
the combination of a good general education, business experience in 
more than one field, and supplementary training in some academic 



vi Preface 

discipline. These broadly equipped persons would be well prepared 
to attack today's massive corporate problems. It is true that many 
companies send their executives back to school for short term re- 
freshers or orientation in new fields. This activity undoubtedly in- 
creases managerial capability. However, the absence of general theo- 
ries in business penalizes executives who would like to find a frame 
of reference through which they could do more effective problem 
analysis and decision making. 

The superficially similar characteristics of systems have naturally 
led researchers to apply the tools of scientific disciplines in less dis- 
ciplined fields of endeavor. Sociology, anthropology, history, eco- 
nomics, and political science are areas of study coming under the 
scrutiny of specialists who have exposure to more than one field of 
knowledge. These specialists are searching for general theories to 
improve the organization of knowledge about a particular area of 
study. For a pertinent example, my recent work in the area of city 
planning 1 has demonstrated the usefulness of the systems approach; 
reaction to this material indicates that systems analysis has some 
usefulness for professional planners. 

My effort to attack the problem of a general systems theory in the 
business management field began in 1956 with the development of 
case material for an extension course to be offered at the University of 
California at Los Angeles. Subsequent to this, the opportunity to test 
the usefulness of a systems approach in business became possible in a 
variety of industrial problems which I encountered as a consultant. 
In addition to providing a useful frame of reference in complex prob- 
lem areas, the systems approach became a powerful analytic tool in 
problem identification and problem solving. 

This book, therefore, has a dual purpose: To contribute to a 
general systems theory in the field of business management, and to 
provide a practical means of understanding and applying the funda- 
mentals of systems analysis in the business environment. In the 
former task, I hope others will join and carry this work forward in 
greater detail; as an early entry in this effort, this book has only be- 
gun to explore the possibilities of a general systems theory for busi- 

l Stanford L. Optner, Report on the Feasibility of Electronic Data Processing in 
City Planning, to the Department of City Planning, City of Los Angeles, 1959. 



Preface vii 

ness. The work of others will increase the usefulness of the ideas I am 
setting forth. 

The initial problem was, of course, to state the thesis. Chapters 1 
through 5 are devoted to this end, concentrating on the elements and 
definitions, the fundamentals, and the method. The thesis has been 
stated both academically and realistically by frequent examples that 
come from the business world. 

Chapters 6 through 9 are devoted to electronic data processing 
equipment. In a relatively few years, computers have caused a major 
revolution in data processing. Wherever data is to be processed, sys- 
tems analysis becomes a major consideration. After the first rush of 
electronic dizziness, 2 it has become evident that the effective design 
of systems is still the fundamental requirement of successful com- 
puter applications. Computers have had a special impact on the 
growth characteristics of industry. The use of electronics has tended 
to force companies toward centralization of data processing, al- 
though decentralization is taking place in manufacturing, engineer- 
ing and marketing areas. 

The intent of Chapter 7 is to give the systems analyst a single 
source where a few examples of computer equipment can be com- 
pared in parallel. (There are several manufacturers whose equip- 
ment is not listed, and their absence from this chapter is not to be 
construed as a condemnation.) This cross section of available equip- 
ment describes many of the characteristics and differences in equip- 
ment and will be especially useful to practicing systems analysts. 

It would be desirable to cover Chapters 1 through 9 parallel with 
the analysis of case studies in Part Two. Each case emphasizes one or 
more situations in which systems analysis contributes toward a more 
incisive understanding of the company's problem. The case study 
method works most effectively in a group. Because cases are explored 
through discussion, participants have an opportunity to draw on the 
range of each other's knowledge as the group discusses important 
issues. Cases have been designed to enable participants to express 
many possible solutions, although the fundamental systems prob- 
lems may not be obvious. Cases must be thoroughly digested prior 
to discussion. Cases have been designed for teaching purposes, and 

2 A Business Week Special Report: Computers, McGraw-Hill Publishing Co., Inc., 
New York, 1958. (June 21, 1958 issue of Business Week.) 



viii Preface 

do not deliberately indicate the policy or practice of any company, 
corporation, or individual. 

Throughout the preparation of text and finished copy, I have had 
the assistance and guidance of many people. Among these, I would 
like to mention my friend and colleague Jack K. Weinstock who con- 
tributed substantially to many of the basic ideas in this book and as- 
sisted in editing the manuscript. I am also indebted to Professor Wal- 
lace J. Richardson, Department of Industrial Engineering, Lehigh 
University, for his continued interest in this material during the 
years of its formulation, providing constructive review and critical 
comment. Finally, I must acknowledge the work of Michael Sunder- 
meier of Prentice-Hall in editing, proof reading, and the hundred 
and one functions that take place behind the scenes to produce a 
readable text. 

s. l. o. 



Contents 



PART ONE: SYSTEMS ANALYSIS 

1. The systems concepts in business 3 

Structured systems 3 

Incompletely structured systems 6 

The organization of a system 9 

The factory as a system 12 

The factory subsystems 15 

Integration in business systems 17 

2. Describing the system under study 20 

The boundary concept 20 

Boundaries as isomorphic systems 25 

Filtering input and feedback 26 

Illustrating systems 28 

The steps in systems design 30 
Investigation • Interviewing • Hypoth- 
esis • Implementation 

3. Fundamentals of system design 41 

Alternatives in system design: one-for-one system change- 
over 42 
Designing a new system 43 

Benchmark problems • Mechanization in 

system design • The computer as processor 

Side effects 48 

Optimal system design 50 

Systems purchasers • Criteria and meas- 



Contents 

ures of effectiveness • Inventory status 
as criteria for a scheduling system 

4. Postulating data-processing systems 60 

Abstraction in the business world 60 

Design of feedback loops 63 

Design of control mechanisms 66 

5. Preparing for the systems study 71 

The assignment 71 

Defining the problem area and boundaries 74 

Priority of effort and schedule 
Method of operation 80 

Suggested checklists helpful in setting up the assignment 82 

/. Systems review checklist • II. Internal 
review checklist 
Flow charting 86 

Staffing the project 90 

Selling the assignment 

6. Electronic data processing systems 93 

Early data processors 93 

Electronic data processors 95 

Characteristics of electronic equipment 96 

Speed • Automatic operation • Flexibility 
• Decision making 
Organizing to centralize machine data processing 101 

Making a feasibility study 105 

Questions to be answered by a preliminary survey 109 

Outline of how to conduct a feasibility study 112 

I. Business analysis and problem state- 
ment • 77. Data processing system design 

• 777. Equipment evaluation to implement 

a postulated data processing system 
A glossary of useful terms used in electronic data process- 
ing 115 

7. Evaluation of equipment systems 118 

Introduction 118 

Datatron220 119 

General characteristics and information 

• Number system • Instruction system 

• Storage • Input methods • Output 

methods 



Contents xi 

Honeywell 800 125 

General characteristics and information 

• Number system • Instruction system 

• Storage • Input methods • Output 

methods 
IBM 305 Ramac 132 

General characteristics and information 

• Number system • Instruction system 

• Storage • Input methods • Output 

methods 
UNI VAC Solid-State 136 

General characteristics and information 

• Number system • Instruction system 

• Storage • Input methods • Output 

methods 
Summary and conclusions 140 

8. Constructing the system costs 142 

What are savings? 142 

One-time versus recurring costs 146 

Displaceable and non-displaceable costs 148 

The presentation of system cost data 154 

9. Operations research in business !56 

Brief history 156 

Techniques of operations research 158 

Descriptive statistics • Statistical sam- 
ling and inference • Correlation and re- 
gression analysis • Linear programming 

• Factor analysis • Central system anal- 

ysis • Simulation 
Systems analysis and operations research 162 

Operations research and a general systems theory 165 



PART TWO: CASE STUDIES 

10. G. W. Templar and Company (demonstration 

case) 169 

Making the hypothesis before completing 
the investigation 

11. The Lee Company 180 

Lack of boundaries and an over-all sys- 
tems concept 



xii Contents 

12. Marxson and Company 193 

The need for a feasibility study based 
upon systems principles 

13. The International Corporation 204 

Problem isolation and selection of ob- 
jectives 

14. Carlysle, Inc. 209 

Orientation to a loosely defined, complex 
system 

15. The Simpson-East Corporation 218 

Identifying systems problems in an incom- 
pletely structured business environment 

16. Beaver Alliance Aircraft Corporation 227 

Finding the requirement for coarseness or 
fineness in systems design 

17. A. B. Fleet and Company 238 

Introducing technological improvement 

in an incompletely structured municipal 

system 

18. Wesley Engineering, Inc. 246 

Examining isomorphic systems in a multi- 
divisional business 

19. Davis Engineering Company 256 

Isolating corporate objectives and the 
means of supporting them 

References 269 

Index 271 



PART ONE 



SYSTEMS ANALYSIS 









1 



The systems concepts 
in business 



STRUCTURED SYSTEMS 

Systems can be broadly classed as physical or nonphysical. Physical 
systems are a part of our everyday vocabulary; we are accustomed to 
refer to the combined tubing, electrical parts, accumulators, and 
pumps of a hydraulic mechanism as a hydraulic system, and we 
refer to complexes of equipment for a wide range of uses as electri- 
cal systems, broadcasting systems, telephone systems, transporta- 
tions systems, and so on. 

Scientists responsible for the development of physical systems in 
the last 200 years have pursued the part of the complex phenome- 
non which was simple to understand. 1 This emphasis has shifted in 
the last 50 years; many disciplines have begun to attack complex 
phenomena with the goal of determining what, if anything, can be 
formulated from its parts. This shift from analysis to synthesis 
means that somewhat less emphasis is now placed upon studying 
individual systems and more is placed upon predicting how a num- 
ber of combined systems will function as a unit. 

Experiments with larger, complex systems, at first unacceptable, 
have now become an important facet of scientific investigation. 
Such study has led to the hypothesis of a closed system. The closed 
system can be defined as one which is free of variation or disturb- 
ance. One of the ways to study such systems is through the concept 
of the "black box." The experimenter conceives of a simple ma- 
chine into which he will introduce known inputs and obtain certain 
resultant outputs. These are noted in a format and completeness 

1 W. Ross Ashby, Yearbook of the Society for General Systems Research (Ann Arbor: 
University of Michigan, Mental Health Research Institute, 1958). 



4 The systems concepts in business 

that will describe only those results which are directly observable. 
The problem for the experimenter is then to 

(1) deduce the contents of the black box; 

(2) determine what was not deducible; 

(3) test the rules under which the experiment was performed, to 
determine their validity or need for modification and further 
experiments; 

(4) determine the limitations which were placed on the experi- 
menter; 

(5) determine the usefulness of the output. 

The record of experiments would thus contain the following: 

(1) the inputs which were used; 

(2) the outputs which were recorded; 

(3) the variations which were observed. 

The experimenter, in seeking to learn something of the black box, 
examines his output in search of some force which is operating. 
This makes itself known through the statistical arrangements of 
output which he is free to rearrange, providing that, in doing so, he 
does not alter or introduce something which was not already in the 
record. 

The goal in designing such experiments is to produce a condition 
which is non-varying in time. This results in a controlled or tight 
system. Ashby refers to these systems as information-tight since 
they are designed to admit no disturbance at input. 2 Such systems 
would thus have no gain, only decay while the process is in opera- 
tion. Since the capacity of the black box is "finite," the frequency 
and volume of data to be introduced determines the amount of 
decay (data deterioration), which, in the absence of gain, can be 
measured through observation. These invariant systems have ma- 
chine-like characteristics and are highly predictable. 

Highly predictable systems of the type described, are typical of 
physical systems. These systems are structured, or designed, to 
operate in non-variant, highly predictable ways. Because system 
disturbance may be known to occur, the black box (the hydraulic 
system, the electrical system, and so on) is designed to function 
within statistically predictable limits. Black box components are 

2 W. Ross Ashby, op. cit. 



The systems concepts in business 5 

chosen with the same rigor, taking into account factors such as 
fatigue, stress, and reliability. 

A pre- World War II characteristic of complexes requiring many 
component systems was the relatively independent design of each. 
Components were operationally compatible but design was not com- 
pletely integrated with associated hardware. Each component sys- 
tem (called a subsystem) was designed close to some optimum (most 
favorable condition) in terms of the current level of technology. 
Thus, each subsystem might function adequately in terms of its 
own specific "mission"; but the end product in which components 
were installed fell far short of the optimum goal. 

The design and development of ballistic missiles were projects 
requiring the use of complex subsystems. The scientists charged 
with the development of these projects required a new packaging 
concept. In weapons development, with hundreds of breakthrough 
requirements, technical staffs recognized the need for new propul- 
sion techniques, new guidance mechanisms, new fuels, stabilizing 
equipment, armament, and countless other things. These require- 
ments demanded, in addition, integration and unity of function in 
one package, a consequent complete specialization of purpose, and a 
heavy emphasis on the reliability and performance of the end prod- 
uct. These requirements were extended, moreover, to include the 
ground support equipment, materiel, and man power necessary to 
maintain and operate the end product. This, in brief, is the 
weapons-system concept of the modern day missile, where the nose 
cone is conceived as a subsystem; propulsion and guidance are sub- 
systems; and the missile, its ground support equipment, materiel, 
and personnel are the system. 

Note the parallel structure of this complex system and the ex- 
perimenter's model. In the former case, the system comprised the 
following: 

(1) the inputs and outputs 

(2) the processor (black box) 

(3) the experimenter 

This man-machine combination was quite manageable in terms of 
size, compared to the missile system. And yet, if one were to deal 
with a subsystem of the missile — or one of the major assemblies 
comprising the subsystem — one would soon be at a level where the 



6 The systems concepts in business 

inputs, processor, and human factors would be more manageable, 
too. Systems of the physical type contain the intrinsic structural 
properties of the experimenter's model. However, the manufacture 
of components or systems admits a very large number of variables, 
mainly in the form of human factors. Some typical human factors 
follow: 

(1) The engineers who design for production 

(2) The materiel personnel who order parts 

(3) The engineers who design tooling and call out manufacturing 
sequences 

(4) The personnel who plan and schedule production; 

(5) The operators who make, assemble, inspect, and test parts 

The experimenter's laboratory model will seldom be designed to 
operate within the wide bounds of the complex subsystem. As near 
as one might come to this concept in the industrial world would be 
the pilot plant which functions as the economic and operational 
model for a future product. The goal in the industrial world would 
be to use the experimenter's model as the means of isolating indi- 
vidual phenomena, whose ultimate use may not be clear. The 
laboratory model might act as the means of eliminating certain 
possible courses of action. Systems which are designed to operate 
with humans will not qualify as closed systems. 

INCOMPLETELY STRUCTURED SYSTEMS 

I will use the term incompletely structured to describe systems of 
the industrial and business world; this tabular presentation will 
qualify the structured and unstructured systems for the present: 

Property Structured Incompletely structured 

1. Input Invariant; no disturbance Variable; many disturbances 

2. Output Predictable; statistically stable Unpredictable; statistically un- 

stable 

3. Processor Machine-like Man, or man-machine 

Incompletely structured systems are of two general types. The man- 
machine system noted above, an example of which was the missile 
system, is one. The other differs from this in that it does not neces- 
sarily contain a physical system. Such nonphysical systems are those 



The systems concepts in business 7 

in which the human being and his man-made institutions execute 
all of the processor's functions. Following are typical functions to be 
performed in nonphysical systems: 

(1) Planning 

(2) Investigation 

(3) Designing, inventing, creating 

(4) Classifying, sorting, calculating, summarizing, recording 3 

(5) Problem solving 

(6) Decision making 

Since there is no model which dictates that investigation should 
or will come first in every human action, in all of these examples, 
there is overlapping and much interchange. There is also no way 
of knowing how efficiently each phase of the above sequence will 
be implemented. Thus, when man becomes the processor of a 
system, his functions will be much more loosely executed than those 
of the machine in the physical system — we do not expect machine- 
like efficiency, reliability or accuracy. Some examples of nonphysi- 
cal systems in which man may be the sole processing agency follow: 

(1) Engineering 

a) Part number assignment 

b) Drawing control 

c) Prototype or production drawing release 

d) Change control 

e) Drawing room methods 

(2) Administration 

a) Wage and salary review 

b) Labor and materiel requirements 

c) Purchase of manufactured parts 

d) Development of time standards 

e) Release of orders for manufacture 

There are many, many areas not listed under Engineering and 
Administration which are examples of incompletely structured non- 
physical systems. There are, also, a large number of similar systems 
in Sales and Manufacturing. Generally, except in highly automated 
companies, the majority of systems must consider human factors. 
There are some additional characteristics of nonphysical systems 

3 T. F. Bradshaw, "Automatic Data Processing Methods," Proceedings, Automatic 
Data Processing Conference, September 8 and 9, 1955 (Boston: Harvard University 
Graduate School of Business Administration, 1956), Chapter One. 



8 The systems concepts in business 

which apply broadly to man's creations, such as business, industry, 
government, law, cities, and so forth; these are listed below: 

(1) The potential of a large variety, quality, and quantity of related 
and unrelated inputs 

(2) A large number of processors of finite capabilities, occasionally 
working at underload or overload capacity 

(3) A large number of anticipated and unanticipated outputs 

(4) Control mechanisms which operate with unequal efficiency, pro- 
viding the opportunity for unequal applications and results 

(5) All actions not measured with equal rigor and seldom reintro- 
duced into the system to improve future performance. 

These characteristics are operational in nature. However, there 
are some additional, general nonphysical systems characteristics 
which are not operational: 

(1) These systems are generally in a qualitative (non-quantitative) 
state of development. 

(2) Abstraction is at a very low level, and the bulk of available data 
is in descriptive, non-numerical form. 

(3) Comparisons between systems reveal large numbers of super- 
ficial similarities (isomorphisms), most of which are poorly de- 
fined, uncatalogued, and whose causal relationships are poorly 
understood. 

(4) The self-organizing (ability to be self-regulating) or homeo- 
static characteristics which occur, seldom appear in systems 
other than those which are highly generalized, and then only as 
mass effect. 

It is now possible to complete the table which had only three ele- 
ments at the outset, and now has five: 





Property 


Structured 


Unstructured 


1. 


Input 


Invariant; no disturbance 


Variable; many disturbances 


2. 


Output 


Predictable; statistically stable 


Unpredictable; statistically un- 
stable 


3. 


Processor 


Machine-like 


Man, or man-machine 


4. 


Control 


Reliability approaches 100 per 
cent 


Wide range of reliability 


5. 


Feedback 


Self-organizing 


Outputs are not automatically 
reintroduced to improve per- 
formance 



The systems concepts in business 9 

There appear to be five system properties, or elements. These are 
the attributes of any on-going process, structured or unstructured. 
It follows that when an on-going process is found, a system may be in 
operation and the analyst can proceed to define the attributes from 
direct observation. 

The electronic computer has been specially designed to execute 
the general purpose data processing requirements of its users. Thus, 
it will be informative to step back, and look at the computer as a 
system. 

THE ORGANIZATION OF A SYSTEM 

Figure 1-1 shows the organization of a computer. Note there are 
three essential areas, input, output and the central processor. The 





1 


CJ 








1 

R [ 

! ^ 




INPUT 


ENTRAL. FKUUuasc 


OUTPUT 






1 


t> 


STORAGE 




.j 








1 


\ 












CONTROL 










t 


1 


r 










ARITHMETIC 
AND LOGIC 

























Fig. 1-1. The organization of a computer. 



combination of input and output peripheral equipment with a 
central processor is referred to as a computer system. Thus, the 



10 The systems concepts in business 

electronic data processing system is based upon at least these three 
requirements: 

(1) A means of getting into the central processor in order to do 
something, 

(2) A means of getting out of the central processor after something 
has been done, 

(3) A means of going about the business of doing something in a 
reliable, automatic way. 

However, the processor must have the ability to reject improper 
raw data, inconsistencies or intermediate results that do not con- 
form to the set of rules under which it has been instructed to oper- 
ate. There must be, as an intrinsic part of this processor, a device 
that monitors its internal operations. This sets up the requirement 
for a fourth system element: 

(4) A means of monitoring the processor so it will operate in a pre- 
scribed way. 

There is still another system element, however, which hasn't been 
isolated. It is like element number one, in that it is a kind of input, 
but it is unlike number one because it is generated by the action of 
number two. This element is also related to number four in that it 
reports on the behaviour of the system. How r ever, instead of moni- 
toring the processor, element five reports output, bringing the re- 
sults of the process back into the system so that corrections in future 
input can be incorporated. Thus, we describe the final system ele- 
ment as: 

(5) A means of monitoring output, delivering the results of opera- 
tion back into the system as input, to correct future output. 

The computer is a physical (man-machine) system. It achieves its 
operational systems characteristics by its design. The computer is 
composed of transistors, diodes, switches, wires, tubes, and so on, 
which when energized will perform, under the direction of its 
operators, those functions for which it was designed. 

Having referred to the missile as a weapons system, it will be use- 
ful to apply the recognized system elements, and see what they re- 
veal. The missile in flight will have: 



The systems concepts in business 11 

(1) A set of Inputs, coded signals, called a program, which will tell 
the missile what to do. 

(2) A set of Outputs, the speed and direction in which the missile 
is travelling, which are the result of its program. 

(3) A Processor, a computer or similar device which accepts instruc- 
tions, processes them, and is the operating unit on which all 
system elements work. 

(4) One or more Controls, the built-in program that has been de- 
signed to keep the missile on its course, and applies the rules 
under which the on-going process will take place. (There are 
many other controls, such as configuration, reliability, and so 
on.) 

(5) A Feedback, the transmission of output data as another input, 
to correct any discrepancy between what the missile is doing, 
and what it should be doing. 

A convenient way to look at this set of processes is borrowed from 
the way in which the computer system is organized (see Figure 1—2). 

CONTROL 



INPUT- 



PROCESSOR 



>. OUTPUT 



FEEDBACK 



Fig. 1-2. 



There will be frequent reference to this arrangement of system 
elements in the future. It will sometimes be called a module to 
emphasize how it reappears at many levels in the business organiza- 
tion. This module will be used to describe the attributes of a very 
elemental subsystem or a very complex, high level subsystem. It 
makes no exception of the type of data being processed or the type 
of processor being utilized. It provides a way of looking at the func- 
tional relationships which exist in any on-going process. As such, it 
becomes valuable as an objective standard. Since we know every 
system has certain elements, it follows that we can look for the 
system elements when we suspect there may be a business system in 
operation. 



12 The systems concepts in business 

The missile in flight to the target was seen as a system. The systems 
engineer must learn in orderly steps of analysis, to look at business 
processes as systems. It is with this general frame of reference we 
can now begin to employ the systems approach in the business and 
industrial world. 

THE FACTORY AS A SYSTEM 

The first requirement is to look at an on-going business, out of the 
eyes of a systems engineer. The business enterprise must be viewed 
as a system. So the first logical question to ask is what are the out- 
puts, always starting with output since it establishes the purpose for 
which the system exists. This question is easily answered, since the 
output of a manufacturing company, for instance, would be a sal- 
able product and a state of profit or loss which is the company's 
measure of effectiveness. In this example of the manufacturing 
company, if the outputs are a salable product and profit or loss, what 
are the necessary inputs? They may be numerous, but they can be 
generally stated as direct and indirect labor, direct and indirect 
materials and capital, in the form of cash, land, buildings, and 
equipment. 

Note that the factory is the system under study, and that its place 
in the module is as the processor. In all cases, the system under study 
will be the processor. Only in a factory of a special configuration 
will it be possible to provide the functional relationships (processes) 
which will change the input from its original state into the required 
output. It will be extremely important to determine when the 
processor can or can not produce the output. This may happen 
either because the inputs are unsuited to the process, or because the 
processor is in some way unsuited to the inputs. This could also 
reflect inadequate control or feedback, as well. Unprofitable opera- 
tion can be characterized as a business state that exists when system 
elements do not combine in an optimum way. The converse may 
also be stated. Profitable operations will exist when the business 
system elements combine in a way which approaches an optimum. 

What are the controls which operate on the factory system? Cer- 
tainly there are legal requirements to consider. For instance, in the 
case of a food products manufacturer, the Pure Food and Drug laws 



The systems concepts in business 13 

would apply. There are additional controls in the form of a mini- 
mum hourly wage or other federal, state, and municipal restrictions. 
These controls may be called external to the system in the sense that 
they apply with more or less equal rigor to all factory systems. Such 
external controls must be considered within the system boundaries 
if they apply to the system under study, and their use as a constraint 
must be reflected in the operations of the factory. 

Controls can be considered internal to the system when they 
monitor the way in which operations are to be conducted. Any par- 
ticular factory system under study will have policies, organization, 
and a plan under which it operates. These can be quite specific and 
have the ability to monitor the factory operation. At this point, we 
must assume the controls are numerous and precise enough to keep 
the system in profitable operation, something that will be examined 
in more detail at a later point. 

Finally, we must determine whether or not there are mechanisms 
at work that operate like feedback in the missile system. Certainly 
one of the results of delivering a salable product to a customer is the 
reorder reaction. If the product has sold, it will be reordered; if it 
has not sold, there is less likelihood that it will be a candidate for 
reorder. The sum of the reactions of all customers to the initial 
shipments of a new product line are a very useful feedback. The 
factory may stop manufacturing slow moving items and concentrate 
on faster selling numbers. This may be a response to style, competi- 
tion, climate, economics, or any other of a number of conditions or 
combinations of conditions. 

What about the factory which ships parts that pass its final in- 
spection but are rejected at the customer's receiving inspection? 
Here the feedback may be that parts are unacceptable because of 
the following conditions: 

(1) Made from the wrong material 

(2) Shipped too late; order had been cancelled 

(3) Shipped too early; cannot use 

(4) Out of tolerance 

(5) Intended for another plant of this division 

(6) Quantity greater (or less) than ordered 

(7) Shipped with inadequate paperwork 



14 The systems concepts in business 

Thus, it is clear that there are feedback mechanisms at work — a lot 
of them. The problem for the systems engineer is to design the 
system to take advantage of all the system elements. In doing this, he 
must assess which of the possible candidates for the systems elements 
are critical, since those which are not critical will not be used. The 
processes by which systems elements are isolated and their im- 
portance in the system operation analyzed will be covered in Chap- 
ters 2 through 5. Economy of system design dictates that there 
should be a balance between all possible system elements and those 
which are important in perpetuating the optimum system opera- 
tion. 

It is possible that a system will be designed with all the elements 
and still not provide optimum output. All system elements must 
be reliable. If, for instance, you are depending on individual pur- 
chasers to return a factory guarantee registration slip which is 
packed in every new appliance carton, you are willing to accept less 
than perfect feedback. If an incentive to return these slips were pro- 
vided, the result might improve. Accuracy is a major requirement 
of effective system design. If a manufacturer plans his shippable in- 
ventory based on the level of store inventory, as many seasonal man- 
ufacturers do, the means by which he derives his knowledge of 
current store inventory can point him to bankruptcy as fast as it 
might place him in the forefront of his competitors. Frequency is 
an important system characteristic, since the updating (making 
current) of records should contain the minimum lag. In one week of 
daily information, a change in trend might be discernable — and 
ultimately provide the clue to a market reaction — but weekly data 
for the same purpose might be inadequate since it would take three 
or four weeks to identify a new trend. Completeness of data is man- 
datory for proper system design. Completeness does not necessarily 
mean that 100 per cent of the available input will be incorporated 
into system operation. Sampling techniques may make it possible 
to reduce input requirements, substantially. Sampling techniques 
must be geared to an individual problem. If the system is to be the 
sensitive device that assists in decision making, it will be essential 
that it return sufficient data to insure a high probability of success. 
These criteria of adequate system design pertain to all system ele- 
ments. The factory as a system is illustrated in Figure 1-3. What 



The systems concepts in business 15 



POLICIES, ORGANIZATION, 
AND PLAN OF OPERATION 



LABOR, MATERIAL, 
AND CAPITAL 



FACTORY 



SALABLE PRODUCT; 
PROFIT OR LOSS 



RESULTS OF SALES 
AND PRODUCT USE 

Fig. 1-3. 

has been said so far is not specific enough to be useful, as yet; 
the model of the factory system is true of most factories. So it is clear 
that, in order to be very useful, the factory will have, first, to be 
restated as a specific factory making a specific product; secondly, it 
will have to be stated as the sum of its most important subsystems. 
Much as the physical components of a missile are interlocked in 
design and function, the subsystems of a business must be intimately 
interrelated. The goal for the systems engineer is to find the means 
of expressing this relationship with the over-all objective of optimiz- 
ing the well-being of the company and the users of the product. 

THE FACTORY SUBSYSTEMS 



Because there is no specific factory under study at this time, we 
will make a general statement of an important subsystem common 
to any manufacturing plant, called Production Control. If Produc- 
tion Control is the processor, what is its over-all objective? Probably 
the overriding objective would be to maintain schedule. The 
schedule is assumed to be an optimal trade-off (exchange of values 
to come closer to a "most favorable" solution) from another sub- 
system called Order Release. This subsystem compromises all of the 
possible shipments with those that can be made within certain cost 
and time (working days available) restrictions. Thus, we are sure at 
the outset that a customer delivery schedule is one of the major 
inputs to our Production Control system. A delivery schedule im- 
plies the existence of a counterpart, closely related, to the produc- 
tion sequence of operation. Thus, we can be reasonably sure that 



16 The systems concepts in business 

the production schedule is a major input to the Production Control 
system. 

The most important material input will be inventory status, tell- 
ing the number of parts in process, in stock, in transit, on order, 
and short for each future manufacturing period. There must be a 
basic labor counterpart for the scheduled inventory status which 
was the primary material input. In the case of labor, it could be a 
labor load (long range schedule of direct labor requirements), or it 
might be a machine load (short range schedule of direct labor re- 
quirements tied to specific machine classes). 

One important ingredient is missing: The technical package that 
tells what operations to do on each part, the sequence of operations, 
the tooling, the machine tools on which operations will be done, the 
dimensions, tolerances, machine finishes, and so on. Now we can 
organize these elements in a modular fashion (see Figure 1-4). Note 



DELIVERY SCHEDULE 
PRODUCTION SCHEDULE 
INVENTORY STATUS 
MACHINE LOAD 
TECHNICAL PACKAGE 



PRODUCTION 
CONTROL 



-►MAINTAIN SCHEDULE 



Fig. 1-4. 



that our system module is incomplete because the control and feed- 
back elements have not been described. Here are some of the im- 
portant controls which should be operating in this major subsystem: 

(1) Organization role of Production Control 

(2) Master Schedule — the company's top level plan of operation 

(3) Management decisions based on periodic reports 

(4) Minimum lot sizes or other rules relating to releasing orders for 
production 

(5) Standard practices, company policies, rules, and so on. 

There are other controls working in this general system, and it 
would be essential to find them in a real-life situation. 

Now for the feedback: One of the most valuable is the progress 
report which shows the status of work-in-process. Like the missile in 
flight which signals its flight path to tell where it is, the progress 
report does the same thing for the factory operation. Another docu- 



The systems concepts in business 17 

ment which tells more about the output is the performance report. 
Here, the standard hours for the parts which have been completed 
are compared to the actual hours expended, as a measure of effi- 
ciency and successful scheduling. The actual-hours feedback tells 
the number of hours which can be removed from the machine load 
when the updating process takes place. The budget fills the same 
role in keeping management advised on the state of dollar expendi- 
tures. 



INTEGRATION IN BUSINESS SYSTEMS 

There are five inputs in the model of the Production Control sub- 
system. The data and reports used as input to the Production Con- 
trol subsystem, were outputs of other, more simple subsystems. 
Properly organized, the outputs would be constructed to integrate 
with the higher-order subsystems as inputs. Integration between 
subsystems is essential, since in practice it gives the following re- 
sults: 

(1) Reports are used for many purposes; special purpose reports 
are at a minimum. 

(2) The need for reports is well established by their use in higher 
order subsystems. 

(3) The timing, frequency, and contents have been specially de- 
signed to energize (activate) next higher order subsystems. 

(4) The volume of data is closely related to the minimum require- 
ments to transmit all information essential for higher subsys- 
tem decisions, but not in excess of this amount. Simultaneous 
updating of files affected by one input becomes possible. 

Integration in subsystem design is the key concept of systems 
analysis in business. Integration of systems postulates the trade-off 
between the functional requirements of a subsystem with its im- 
mediately related subsystems. Integration in this sense also means 
the intercompatibility of subsystem design. Each subsystem be- 
comes a black box when viewed by the systems engineer. Using the 
data at his command, the systems analyst observes the subsystem 
operation, looking at inputs, outputs, and the processing device (the 
black box) to determine whether or not each subsystem is making 



18 The systems concepts in business 

the desired contribution to the over-all system (the factory; require- 
ments. The systems analyst must deal with the known facts to de- 
duce the need for redesign of the system. Since this is not a closed, 
physical system, he will deal with more inputs, outputs, and a more 
loosely defined processor than that of the experimenter working 
under laboratory conditions. The statement of systems boundaries 
indicates how the systems analyst will deal with the design of an 
incompletely structured business system. 

The dependence of higher order subsystems on more elementary, 
lower order subsystems for input must be reminiscent of the original 
description of the missile system (see page 5). Lower order sub- 
systems are, in turn, dependent upon still simpler, more funda- 
mental subsystems for input. The factory system is actually a tightly 
(or loosely) interlocked collection of subsystems. While its needs 
are more general and of wider range than that of the physical system, 
we can take some valuable clues from the analogy. The integration 
of physical subsystems should accent utility (need), economy, and 
unity of purpose. All of the factory subsystems are man or man- 
machine systems. The size of the factory and its product produc- 
tion rate and other factors will determine the exact balance of men 
and machines. The dependence on manual functions in business 
brings about the symptoms that have been associated with the un- 
structured system. 

The first requirement in assessing a business system is to state 
the problem. This means that existing operations must be seen as 
systems and the systems engineer must decide the systems elements 
which are present and missing. This definition of the system under 
study would have us begin by looking at the whole or the largest 
part of the whole, whichever is more manageable and practical. 
Understanding the existing system operation will lead to the ability 
to state the systems requirements from which the design of the 
improved system will follow. 

Having defined the properties of unstructured systems, it is ap- 
propriate to ask whether or not they can ever be better structured 
than they are in the best managed company. Parkinson's Law would 
tend to show that the mere existence of a business with its attendant, 
gradual personnel enlargement will invariably open the possibility 
of systems analysis to restore or improve the level of operations. As 



The systems concepts in business 19 

business grows, it will never be less complex in its goals and means 
of achieving them, only more complex. Both operations research 
and electronic data processing have had some impact on the need 
to structure business systems more efficiently. However, both have 
some limitations in the universality of their application and in the 
attendant costs. Small business enterprises make little use of these 
tools, yet their problems are relatively the same. 

The usefulness of systems analysis is that it provides a much 
needed, objective frame of reference. Concepts such as systems inte- 
gration are invaluable as a tool in problem solving. The systems 
approach supplies a number of principles such as the one relating to 
the system elements. These principles will make analysis and prob- 
lem solving easier and more effective because they fill the need of 
providing basic and elemental ways of describing all business opera- 
tions, regardless of complexity or type. Systems analysis in business 
is built on the premise that there are striking similarities between 
the way physical systems and business systems function. The princi- 
ples of systems engineering suggest that we can learn something 
about the way an effective business system should operate by using 
as a tool the analogy to a logically designed electronic system. Sys- 
tems engineering requires that business operations consist of a set 
of intimately related subsystems. The integration of subsystems is 
the key to effective and economic operation. 



2 



Describing the system 
under study 



THE BOUNDARY CONCEPT 

The system processor is deceptive in its simplicity. It is clear that a 
man or a machine can be called a processor for the purpose of ana- 
lyzing a problem; but when the processor is the Production Control 
subsystem, it is not nearly so simple to say where production control 
starts and stops. Nor is it easy to say just what subsystems will be 
included, and what subsystems will be excluded. This is a matter of 
how the organization defines Production Control, and may be so 
subjective that it merely reflects an individual's opinion of what 
Production Control should be doing. The systems engineer defines 
the system under study by stating its boundaries. The use of a 
boundary concept makes it possible to define any on-going (non- 
static) process as a system. It further enables the systems analyst to 
look at the problem as a whole, and set up the framework for later 
looking at the parts (the subsystems) in something close to their 
correct relationship. 

There are some statements that could be made about all produc- 
tion control systems. However, such statements would have to be 
very general and therefore of marginal value in solving a specific 
operational problem. In any real-life situation, objective recording 
of observable phenomena is requisite to understanding the process. 
A statement of systems boundaries is dependent upon an ability to 
define the system under study. This becomes the goal of initial in- 
vestigation into the nature of the system. In attacking the system 
under study, the engineer makes only one a priori conclusion: The 
systems requirements will dictate how the system should be de- 

20 



Describing the system under study 21 

signed. A knowledge of the system requirements will follow from an 
understanding of the following: 

(1) The activities which are associated with the on-going process 

(2) The inputs which are processed in the system activity 

(3) The outputs which are obtained as a result of system processing 

(4) The way in which the on-going process is controlled 

(5) The errors, deviations, and exceptions which have been marked 
as system malfunctions. Included in this category are the fol- 
lowing: 

a) System malfunctions in time: that is, the time element nec- 
essary to introduce input or obtain output, feedback, or con- 
trol inhibits the function of the system or its next in line sub- 
system, 

b) Cost malfunctions: that is, the costs to introduce input or 
obtain output, feedback, or control are greater than they 
were in the past; or the costs to sustain the system are intui- 
tively determined to be higher than the value of the system 
operation itself. 

A statement of the boundaries will, by determining the content of 
the system, place limits on the system to be studied. This is necessary 
since an ability to define the adjacent areas of a system will empha- 
size the integration problem. It will have the additional value of 
concentrating investigation in areas where evidence pertinent to 
the solution of the problem is available. 

A boundary in the systems' sense restricts the scope of a problem 
to a size commensurate with the cost or time available for solution 
and the amount of detail necessary to understand the process. It 
is possible to view Production Control in many ways and change 
the boundaries accordingly. Management's view would tend to re- 
late Production Control to the organization as a whole, and Man- 
agement would tend not to see it as an operating unit filling a 
variety of minute, day-to-day needs. The operators or users of Pro- 
duction Control services would tend to think in non-organizational 
terms, perhaps only looking at one or two special functions per- 
formed for them alone. These statements are not necessarily contra- 
dictory — they only emphasize the range of definition possible for 
the processor. Thus, it is entirely right that the system under study 



22 Describing the system under study 

be accurately described in terms of the role the processor plays, in 
a specific instance (see systems purchasers, Chapter 3). 

A boundary of a subsystem will create a problem of manageable 
size. Systems which include many subsystems will, in proportion 
to the complexity of the organization, tend to be less easily managed 
or controlled. The ability to look at systems and subsystems with the 
same set of requisites, improves the ability of the analyst to main- 
tain uniformity and consistency in the statement of objectives, 
criteria, and assumptions. This ability improves performance in 
problem recognition, problem identification, and fact gathering for 
problem solving. 

The statement of boundaries will assist in determining whether 
an output can be produced, given either the processor or the input. 
There must be additional conditions which specify something about 
the behaviour of the missing elements, as well. For instance, if you 
are told that one measure of an effective Production Control would 
be a reduction in scrap — would you agree? This module is illus- 
trated in Figure 2-1. One is quickly and inescapably led to the 



PRODUCTION 
CONTROL 



REDUCTION 
->»0F SCRAP? 



Fig. 2-1. 



conclusion that there actually are some things one could do to 
achieve this desirable goal. Some of the possibilities are outlined 
below: 

(1) Input. Improve quality of technical package, reducing scrap 
caused by 

a) inadequate drawings 

b) inappropriate tooling 

c) poor materials 

d) improper cutting tools 

e) improper feed, speed, depth-of-cut 



Describing the system under study 23 

(2) Control. Provide special rules and procedures designed to re- 
duce scrap caused by 

a) inadequate supervision or in-process inspection 

b) wrong man working on the job 

c) improper operation of machine tools, hand tools, hand 
checking devices. 

(3) Feedback. Find means of relating actual quantities of scrap to 
causes of scrap, bringing appropriate changes into the system 
to modify past scrap-producing practices by 

a) increasing standard hours allowance: operators generate 
large amounts of scrap attempting to meet or beat standards 

b) increasing lot size: scrap per cent shrinks as large quantities 
are processed; bulk of scrap occurs in first fifty pieces 

c) not permitting operators to make their own setups: substi- 
tute specially trained set-up men who will do more precise 
work. 

However, the major factor in producing scrap has not been con- 
sidered because this is actually the wrong subsystem. The module 
that should have been analyzed is illustrated in Figure 2-2. The 



TECHNICAL DATA, 

INSPECTION RULES, 

FACTORY PROCEDURES, 

SUPERVISION 



GOOD PARTS 
FROM PREVIOUS 
OPERATION 



OPERATOR 
AND MACHINE 



GOOD PARTS 
■► FOR NEXT 
OPERATION 



PARTS REWORKABLE 

Fig. 2-2. 

rule is to look at the processor most directly concerned with the 
problem. In this case, the systems analyst would study the fabrica- 
tion process in the attempt to reduce scrap. Human error makes 
more scrap than machine error. Technical data, factory procedures, 
inspection, etc., are seen as controls over the on-going production 
process, not input (compare Figures 2-1 and 2-2). If the goal is to 
reduce scrap, one way is to apply controls and feedbacks, taking the 
part of the output which is not good, and rework it. This module 
more faithfully reflects how to achieve the desired goal. 



24 Describing the system under study 

It is now possible to deal with the problem of whether or not re- 
duction in scrap can be considered a primary objective of a produc- 
tion control system. The answer is, no. There are some more or less 
direct, and many indirect effects of production control actions 
which tend to reduce scrap. However, reduction in scrap is really 
more closely related to the machine, the machine operator, inspec- 
tion, the technical package, etc. The system module must be ana- 
lyzed to determine whether or not the system elements are actually 
related in the way they have been postulated. It is especially im- 
portant to test the processor to see if it is capable of producing the 
output. The processor must define the boundaries within which the 
on-going process will take place. 

The ability to find the correct subsystem or to identify the proper 
systems elements, emphasizes the art in systems analysis. As in any 
other field, the ability to perform efficiently is to some degree a 
result of the participant's training, experience and ability. Thus, 
the business-oriented systems analyst can make a major contribu- 
tion because of his knowledge of business operations, generally. 
Knowledge of or experience in the system under study will speed 
the analysis and identification of systems elements. In the absence 
of such knowledge or experience, the systems analyst can rely upon a 
knowledge of the properties of systems to aid him. An under- 
standing of systems analysis will never replace a knowledge of the 
problem under study, but it will provide an objective frame of 
reference in which investigation and hypotheses can logically occur. 

The system boundaries can now be defined for data processing 
purposes as the following: 

(1) A human: in this case, the work performed will be done manu- 
ally. 

(2) An electro-mechanical, man-machine combination: in this case 
the work would be performed by men in conjunction with 
punched card (tabulating) equipment, bookkeeping machines, 
and so on. 

(3) An electronic, man-machine combination: in this case, the work 
would be performed by men in conjunction with a computer 
and its associated complement of electro-mechanical equip- 
ment organized as a computer system to achieve certain pre- 
determined ends. 



Describing the system under study 25 
The system processor can be denned for all other purposes as: 

(1) Man-made objects or institutions: in this case, the system may 
be an inanimate object which has an on-going process which 
can be readily described by its system elements. Examples of 
this would be cities, government, and so on. 

(2) Environmental: in this case, the system might be the topog- 
raphy of the land, the watershed and water uses, the land use, 
and so on. 



BOUNDARIES OF ISOMORPHIC SYSTEMS 

The isomorphic (superficially similar) nature of systems can be 
restated for subsystems such as Production Control. There are 
superficial resemblances among Production Control systems. Some 
multiplant manufacturers insist on the uniformity of systems design 
in their operating units, with some variability in the success of 
individual applications. The outstanding issue is to determine the 
conditions under which similar but not identical systems require- 
ments may be adapted to operate effectively with the same system 
design. These are the necessary conditions: 

(1) When both systems have the same "mission" or reasons for ex- 
istence. Example: A machine shop and a sheet metal shop em- 
ploy about the same number of people, operate as job shops, 
generate a comparable dollar volume, and organize their fac- 
tory operations around service departments of the same type 
and size. If we look at these two, independent examples as sys- 
tems, we might come to the conclusion that, despite the differ- 
ences in salable product, both the business objective and the 
means to achieve it have been organized under the same bound- 
aries. In this circumstance, an identical system design might be 
adopted with little or no change. 

(2) When both subsystems have highly specialized data processing 
requirements. Example: A dress manufacturer and a furniture 
manufacturer employ about the same number of people. In the 
preparation of their payroll and associated quarterly tax re- 
turns it would be very likely that both could utilize the identi- 
cal data processing system. 

(3) When each subsystem is extremely simple in its processing role 
and is designed to receive input with no variation. Example: A 



26 Describing the system under study 

retailer of dry goods and a wholesaler of sports equipment re- 
order their stock when the stock level reaches a predetermined 
point. If the number of items to be controlled is approximately 
the same, it is conceivable that both companies could use the 
same number of personnel, the same system and perhaps the 
identical forms to record inventory facts. 

(4) When control is outside the system, and operating on output, 
which becomes input to an adjacent system. Example: See Fig- 
ure 2-3. 



CONTROL 



CONTROL 



»UT — r*. 


PROCESSOR 


. . ^ nirmiY 


INPUT 




REVIEW OF 
PREVIOUS 
SYSTEM 
OUTPUT 




i ' 


k 




























FEEDBACK 







COMPLETED 
REVIEW OF 
.PREVIOUS 
SYSTEM 
OUTPUT 



CORRECTIONS TO PREVIOUS SYSTEM OUTPUT 

Fig. 2-3. 

(5) When decision making is not included in the system operation 
(see item 4). 

(6) When both systems operate well within time restrictions that 
preclude unequal deterioration or decay of data. Example: 
Two divisions engaged in different manufacturing activities are 
required to provide corporate headquarters with reports of 
monthly sales. All reports are due the fifth working day after 
the close of the month. Systems for generating data on monthly 
sales could be identical if they supplied the requisite data 
within the time restriction. 

These conditions restrict greatly the universality of any one business 
system — and rightly so, since each organization has objectives which 
are somewhat different than any other organization. 

FILTERING INPUT AND FEEDBACK 

Even under circumstances where identical products are produced 
by the same manufacturer in different locales, environment acts as a 
filter altering to a considerable extent the nature of the system. 
Environment acts as a filter by allowing only certain input and 









Describing the system under study 27 

processes, eliminating others. For instance, the labor pool in south- 
eastern United States is quite different from that in New England. 
The wage earner of the Pacific Northwest is quite different from 
his counterpart in the Southwest. Similarly, environment has a 
marked effect on the availability of the resources normally provided 
by capital. For instance, Tucson, Arizona, which has few factory 
buildings or facilities, must provide space to encourage immigration 
of new business. Los Angeles, with a wealth of facilities available, 
has passed the point where the Air Force feels it is desirable to con- 
tinue to concentrate certain industry in its environs. Climate is, of 
course, a major environmental filter, as well. 

The filter is a useful concept and is defined as a man, man-made, 
or environmental factor which consciously or by its state of being, 
acts to admit certain system elements to the system process keeping 
others out. In the area of systems design, it would be desirable to 
have filters which allowed the entry of all valid and valuable input, 
automatically rejecting invalid or valueless data. A way to show the 
filter in the systems module is indicated in Figure 2-4. As can be 

CONTROL 



INPUT- 



PROCESSOR 



FEEDBACK 



-►OUTPUT 



FILTER 



Fig. 2-4. 



seen, the postulation (necessary existence) of a filter in a system 
actually infers certain subsystems which cut across input and feed- 
back, but which may lie outside or inside the system. The system 
processor acts as a filtering mechanism as well, producing only those 
outputs which can be created from specified inputs. Likewise, cer- 
tain operations, such as one called edit-input (or edit-output), are 
filtering- type operations. Manual systems generally call for more 



28 Describing the system under study 

filtering-type operations than those which are mechanized or auto- 
mated. In stating the boundaries under which a system will operate, 
the ability to recognize the existence or absence of filtering opera- 
tions, will be necessary. 

It is now more clear why systems can follow no stereotype but are 
at best isomorphic. A host of environmental factors work against 
uniformity. Differences in management objectives are reflected in 
organization and operations. The design of a system is the solution 
to a problem of a specific management. For the systems analyst, the 
solution will arise out of the existing process and a knowledge of 
the systems requirements. 

ILLUSTRATING SYSTEMS 

The systems diagrams of Chapter 1 are typical analytic devices 
for looking at the existing process in a coarse way. They are equally 
suitable for looking at details of the process when the problem is to 
analyze the existence or nonexistence of important system elements. 
Systems diagrams are different from other types of flow charting 
techniques in that they accent the use of the systems elements as the 
important tool in system analysis; they do not replace flow charts, 
which are used for other purposes. These are the things the systems 
module illustrates when it is used as an analytic tool in investigating 
the existing system: 

(1) Identify the system under study (processor). 

(2) Identify the purpose for which the system exists (output). 

(3) Identify the ingredients (input) whose functional relationships 
can be arranged to produce the required end result. 

(4) Show the existence or nonexistence of mechanisms whose pur- 
pose is to maintain reliability, accuracy and other desirable op- 
erational attributes (controls). 

(5) Show the existence or nonexistence of mechanisms to correct 
malfunctioning output (feedback). 

The basic diagramming technique is to assist in identifying systems 
elements (see Figure 2-5). The second diagramming technique is 
to demonstrate the inter-dependent nature of subsystems, when the 
output of the first system becomes the input to its next in line sub- 



Describing the system under study 29 

CONTROL 



INPUT 



CONTROL 



PROCESSOR 



•FEEDBACK - 

Fig. 2-5. 



>» OUTPUT 



CONTROL 



INPUT- 



PROCESSOR 



FEEDBACK- 



•OUTPUT • INPUT- 



PROCESSOR 



FEEDBACK • 



-*- OUTPUT 



Fig. 2-6. 

system without alteration (see Figure 2-6). Modules so related em- 
phasize the integration aspect of data processing. The actual area of 
integration is where the output can be used as input elsewhere 
without alteration. A second aspect of integration is shown in the 
next diagram (Figure 2-7), where, 

I a — Input and output to subsystems P ab and P ac 

P ab — Processor for subsystem ab 

O b = Output h of subsystem P ab 

F ab = Feedback from O b , subsystem P ab 

C ab = Control over subsystem P ab 

P ac = Processor for subsystem ac 

O c = Output c of subsystem P ac 

F ac — Feedback from O c , subsystem P ac 

C ac = Control over subsystem P ac 

Note that the initial input (I a ) is an output of subsystem P ab and is 
reused in subsystem P ac as input, without alteration. This is a typical 
use of systems integration where a computer is the data processing 
device. It is not uncommon in computer applications to have a 
single file of sorted input transactions update or operate against 
half a dozen basic files. If the computer carries the files in external 
storage (magnetic tape units), passing one input across the six tape 



30 Describing the system under study 

units takes place at computer speeds, the bulk of the time being 
consumed in searching for the proper record. Note also that I a can 
be an output of subsystem P ac or P an , if required. Output require- 
ments are a part of the boundaries under which system design takes 
place. 



'ab 



ab 



ab 







\ 




°b 


P ao 




la 


/ 








F ao 







■a 



Fig. 2-7. 



The third diagramming technique is of a coarser nature and in- 
tended to implement the early need to look at the major subsystems 
of an ongoing process, breaking the system into its component sub- 
systems (see Figure 2-8). In this type of diagram, the need to show 



INPUT- 



SUBSYSTEM 
A 




SUBSYSTEM 
B 




SUBSYSTEM 
C 




SUBSYSTEM 
D 









OUTPUT 



Fig. 2-8. 

controls or feedbacks will be entirely optional, depending on the 
usefulness of the detail. The absence of intermediate inputs and 
outputs is intentional. In this method of analysis it is not essential to 
supply all details, if the purpose is to make an over-all breakdown 
to represent the full system. Intermediate inputs and outputs (as 
well as all other systems elements) are invariably implied, however, 
because the definition of a system is an on-going process containing 
all of the five systems elements. 



THE STEPS IN SYSTEMS DESIGN 

To state the boundaries one must make an appraisal of the prob- 
lem prior to actually working on the problem. This appraisal would 



Describing the system under study 31 

normally be limited by how much detail one would have to collect 
to make a careful statement of the problem. The extent of this pre- 
liminary appraisal of the problem would be related, as well, to the 
amount of material required to solve the problem. The danger in 
developing the boundaries is that it is possible to be carried into the 
back alleys of distantly related problems. The need to appraise 
familiar or unfamiliar problem areas in establishing boundaries 
and, subsequently, to investigate problems makes it desirable to 
explore the process by which this should be done (see Figure 2-9). 



^ 



ANALYZE THE PRESENT SYSTEM 



DEVELOP A CONCEPTUAL MODEL 



>- INVESTIGATION 



J 



TEST THE MODEL 



PROPOSE A NEW SYSTEM 



>- HYPOTHESIS 



PILOT INSTALLATION OF THE NEW SYSTEM 



FULL INSTALLATION OF THE NEW SYSTEM 



V- IMPLEMENTATION I 



Fig. 2-9. Steps in system design. 



Investigation 

What is the existing system? This is the first and sometimes the 
most important question the systems analyst can ask; it contains the 
important objectivity requisite to effective investigation. It does not 
presuppose that there is no existing method for coping with a prob- 
lem. The existing method may be malfunctioning or poorly con- 



32 Describing the system under study 

ceived, but these are precisely the symptoms the investigator must 
understand in order to be able to document the problem carefully. 
How does one determine the nature of the existing system} This 
is a period of intensive data collection and interviewing. Basically 
there are two types of informants: Those of the written word and 
those of the spoken word. Both must be used. The written materials 
which are pertinent might take the following forms: 

(1) Examples of inputs or outputs. 

(2) Examples of file records. 

(3) Actual examples of system malfunction. 

(4) Memoranda, letters, and notes which may be pertinent. 

(5) Other reports or commentaries indicating previous study of or 
attention to the same problem. 

The guiding principle is to sample everything of value. Take ad- 
vantage of every type of existing information. Naturally, this can 
go too far if discretion is not used. It is reasonable to assume, how- 
ever, that more raw material is better than less — it is always easy to 
cast aside unusable data. It is not desirable to have to hold up a part 
of the problem for an extra period of data collection. 

The spoken word is at least as valuable. The organization imposes 
certain restraints that make it very likely that many things can be 
learned by word of mouth that will never be found on paper. Thus,, 
there are good reasons for interviewing the people most affected by 
the problem under investigation. But personal interviews can be- 
come confused, redundant, and time-consuming. Worse, they can 
serve the purpose of putting the investigator on the wrong track. 
The position and personality of the person being interviewed can 
inject a pronounced bias to an interview. The hazards of interview- 
ing only make it more necessary to plan each contact in advance and 
fit the material collected by each contact into the pattern which 
begins to emerge early in investigation. 

Interviewing can take the form of an opinion poll when im- 
properly handled. Generally, the investigator wants facts, not opin- 
ions. However, if opinions which may be necessary or desirable are 
solicited, they must be treated as such. The following gives some 
general rules on the conduct of an interview. 



Describing the system under study 33 

Interviewing 

The role of the interviewer is to obtain specific information. He 
will have organization precedents to guide him in his task. He will 
attempt to control the interviewing process without cutting himself 
off from vital information by obstructing an important channel of 
questioning. He will be forced to make decisions as the interview 
proceeds as to what line of questioning is valuable, and what infor- 
mation is relevant — what data he can get elsewhere with higher 
reliability, and how much of the data being collected will require 
confirmation before being used. 

A workable interview technique gets the problem in narrative 
form first, and then follows the leads that come out of the narrative 
to ask specific questions. The interview should start with mutual 
recognition of why the problem is being investigated. The material 
to be covered must be preplanned with the idea of keeping the inter- 
view within bounds. Sometimes it is valuable to compose a state- 
ment which will describe the purpose or object of the interview, and 
to use this as a guide if the problem is very complex. 

Here are some valuable "don'ts" that apply in interviewing: 

(1) Don't interrupt the story to insert your own ideas. 

(2) Don't let the interview get diverted into paths that are obvi- 
ously not pertinent. 

(3) Don't let blanket statements or broad generalizations obscure 
the facts. 

(4) Don't let half-understood problems go; leave the interview with 
a clear concept of the issues. 

(5) Don't be overpowered by the person being interviewed; be sure 
you leave feeling you were the interviewer and not the inter- 
viewed. 

(6) Don't become involved in operational problems or offer solu- 
tions that will distract from the prime purpose of information 
gathering. 

(7) Don't ask questions that can be answered "yes" or "no" with- 
out recognizing that the question sometimes will call for an 
opinion and not a fact. 

Here are some valuable things to do during the process of inter- 
viewing for data collection: 



34 Describing the system under study 

(1) Make personal contact with your subject immediately; keep the 
contact "human." Five minutes of warm-up before you start 
will pay off in good cooperation. 

(2) Describe the assignment briefly so the person being questioned 
will see his part in the total picture; make him feel he is on the 
"inside"; tell him what your role is; invite him to feel the im- 
portance of his role in the program. 

(3) Have an outline of the material you intend to cover and other 
pertinent data with you when you interview; use these as a 
guide to be sure you are getting all the data. 

(4) Make your notes telegraphic but not unintelligible; take care 
in writing so you can read your notes. 

(5) This is a suggested type of indenture in outlining the narrative: 

a) Roman numerals 

b) Capital letters 

c) Numbers 

d) Small letters 

e) Numbers in parentheses 

/) Small letters in parentheses. 

(6) If you are making lists or enumerating data, organize the infor- 
mation for logical and easy understanding on separate sheets., 
noting where to insert or pick up the data. 

(7) If you miss a point and want to insert it, use this method: 

a) Draw an arrow to the point of insertion marked by a capital 
letter (A); in the margin by the arrow, refer to the page 
where the insertion will be found: 

See Insert (3) on Page 9 < > (X) 

b) Keep inserts on separate sheets. 

(8) Be sure to note data such as the names of persons interviewed 
and their titles; head each interview separately for easy identi- 
fication; indicate page number and date. 

(9) Be sure references to flow charts are clear and unmistakable so 
notes and flow charts can be easily associated. 

Soon after the start of the investigation the conceptual model 
will begin to take shape. The conceptual model is the analyst's first 
idea of how to attack the problems of the system and of the way in 
which the system should be redesigned. The conceptual model 
might be a reliable indicator of how the final system will be designed 
— or it might be very unreliable. Inadvertently, the systems engi- 
neer may become involved in solving a problem even at the investi- 



Describing the system under study 35 

gation stage. Unfortunately, it is difficult to avoid trying to find a 
solution during the investigation period; this, however, must be the 
period during which investigation is still the primary goal and the 
systems analyst must make a conscious effort to be a data collector. 

Hypothesis 

This study area can be defined as the period which lies between 
investigation and implementation. It is, despite this somewhat loose 
identification, concerned with two important logical steps, the first 
of which is testing the conceptual model; the second is proposing 
the new system. 

Although Figure 2-9 shows no feedback process, what happens 
during a study is something like that illustrated in Figure 2-10. 
Block 2 is going back to block 1; 3 is returning to 2; and, hopefully 
when the study moves into block 4, there will be a minimum need 



Block l. 



Block 2. 



Block 3. 



ANALYZE THE PRESENT SYSTEM 



M I 



DEVELOP A CONCEPTUAL MODEL 



TEST THE MODEL 



: 



Block 4. 



PROPOSE A NEW SYSTEM 



Z) 



Fig. 2-10. 



36 Describing the system under study 

for more testing. The return arrows which block the proposal of a 
new system reflect the feedback from incomplete boundary defini- 
tion in earlier phases. In other instances, feedback could be of a 
positive type where the system is adequate but simulated operation 
has shown ways in which it can be improved. 

Block 4 is sometimes called postulating the system. This term has 
wide use among engineers doing feasibility studies, because they 
have the need to document and preplan an entire system on paper, 
far in advance of actual use. Thus, their paper plan is known not 
as an actual but as a postulated system, since it presupposes its 
workability without empirical proof. 

The modd of t he system may be in ma ny forms. As a mathe- 
matical expression, it might be readily tested. This is nor always 
possib ler7ts~a busines s system it will meet its most severe test in use. 
But short jrfjthis^if ra r\ receive Q r^] t^t hy praemtinrr it m fog po- 
tential users for their criticism and comment. If paperwork will 
change hands in the operation of the system, representatives of 
departments affected by paper handling should be called in. Com- 
pleting forms, making sample computations, and preparing result- 
ant outputs by simulating the operation of the system, will reveal 
some of the inherent weaknesses. An excellent practice during the 
design of systems would be to bring affected departments into the 
picture as early as possible and to keep them there throughout the 
study. Systems are not designed for individuals and especially not 
for the credit of the systems group. Natural resistance makes it man- 
datory to bring those who must change into contact with those who 
are bringing about the change. If the model of a system is mathe- 
matical, its usefulness in testing will be restricted unless the data 
introduced for processing are empirical. In the absence of data 
scaled to the problem, extensive testing may have to wait until im- 
plementation begins, and the testing will be continuous. 

In a new system, it is Hpsirahlf tr> tpst thp ovpr-all o peration in any 
way feasible. It is also desirablejtojrsf ^p^rxtrfy thp s^frgygt^mg 
that combine to form th e over-all system. For instance, in a ma- 
chine loading subsystem, it would be desirable to check out all 
the paper handling that would have to go on in the factory, in order 
to be certain that the handling of job cards was in keeping with the 



Describing the system under study 37 

system concept. Following are some of the subsystems which it is 
important to test in the design of a machine loading subsystem. To 
the extent that they may be independent, they may be tested inde- 
pendently; to the extent that they are dependent serially, they must 
be tested both independently and serially: 

(1) Check-in for new job assignment 

(2) Pick-up tooling for new job assignment 

(3) Check-out setup after first piece 

(4) Clock-in on production 

(5) In-process spot inspection 

(6) Clock-out on production. 

These subsystems each h ave a marked effect on the success ful opera- 
tion ot tne over-all system. Th eir proper operation will determi ne 
to some exte nt the success of the over- alls ystem. Why test t hem 
"separatel y^ Jbecause each is somewhat complicated in its own right. 
Also, a change in one may have no effect, or may have a marked 
effect on the other — the only way to know this is to un derstand all 
about each subsys tem in d et ail. Finally , it is important tn ^phligh 
subsys tem compatibility. T his is only possible by linking the out- 
puts of subsystems as interdependent units, to see that those which 
happen first in time satisfy all of the requirements of those which 
operate later in time. In the laboratory, the experimenter attempts 
to reproduce reactions a number of times to obtain statistical valida- 
tion. He also does this to be certain that all of the data from his ex- 
periments are properly understood and expressed. Systems analysts 
agree that this is a sound technique and that business will reap a 
high order of effectiveness from its systems if they are designed and 
tested according to this method. 

The problems encountered in testing systems will be complicated 
by the unstructured nature of the problem. Thus, in business sys- 
tems statistically predictable variance will have wider limits than 
that acceptable to the laboratory experimenter. The unstructured 
nature of the problem must be considered in designing the testing 
format to be employed much as it must be considered in the design 
of the system itself. 



38 Describing the system under study 

Implementation 

It is not always possible to have the opportunity to make a pilot 
installation. However, it invariably proves its value when one can 
be arranged. Pilot installations have the obvious advantage of allow- 
ing the system in miniature to operate under real-life conditions. 
This means that defects can be corrected before large scale commit- 
ments are made. 

A good example of a pilot operation would be one which per- 
mitted the trial of a new timekeeping subsystem designed for 1000 
employees working in a factory but did not involve all 1000 em- 
ployees from the outset. The systems engineer in this instance might 
obtain Management's agreement to have one cost center of one 
department (a subdepartment, or the smallest factory unit) trained 
in the use of the new system. This group would be relieved in the 
test period of any responsibility for the old system which is being 
superseded. The group must be large enough to be significant. If 
this factory operates on a multi-shift basis, it would be important to 
determine whether or not the pilot operation should also sample 
second or third shift problems. 

Pilot operations are good in that they concentrate attention on 
the new system. Systems engineers may require many months to 
design a large scale system. If Management becomes impatient, once 
in the pilot stage they will begin to see the fruit of the labor that has 
been expended. The pilot operation of a system thus becomes an 
important goal, both for Management and the systems designer. 
As soon as a subsystem is in pilot operation, it becomes more of a 
reality than it can ever be on paper (see Figure 2-11). Note again 
that problems can occur even at Block 6, forcing alterations in 
system design. It is even conceivable that problems occurring at 
Block 6 might go back to Block 1. In the design of physical systems, 
this is not uncommon. The system designer must work against the 
probability that Block 6 will require the feedback to Block 1. This 
is best done by adequately describing system boundaries and design- 
ing ample testing devices before implementation begins. If a bound- 
ary change occurs due to greater insight obtained in a pilot opera- 
tion, system redesign may be mandatory. Testing the subsystems 
adequately is therefore only a part of the solution to efficient system 



Describing the system under study 39 







p,» 










; n* 




Block 1. 




ANALYZE THE PRESENT SYSTEM 








1 










'* 






Block 2. 




DEVELOP A CONCEPTUAL MODEL 












r 










1 




Block 3. 


TEST THE MODEL 






















g». 

V 






Block 4. 




PROPOSE A NEW SYSTEM 








V 




Block 5. 




PILOT INSTALLATION OF THE SYSTEM 










r 








1 


Block 6. 




FULL INSTALLATION OF THE SYSTEM 




















1 

Fig. 2-U 


' 







implementation. Changes in output requirements may reflect a 
situation where the sum of the optimized subsystems is equal to 
something more or less than the desired optimum system. In this 
sense, the term optimum may be difficult or impossible to achieve, 
and something less than optimum may have to prevail. If the system 
designer can determine within certain limits what the optimum may 
be, this in itself is of considerable help. The system designer must 
act with the idea that his role is to improve the existing process. If 
such an improvement does not achieve the optimum, this does not 
mean that a measurable advantage has not been introduced. 

The concentration of personnel on the problem of a pilot installa- 
tion has the added advantage of newness. New systems are attention 
getters for systems engineers, and they provide an opportunity to 



40 Describing the system under study 

ask and obtain concerted effort from the individuals involved in the 
test. This could have the disadvantage of obtaining exceptional 
performance which is not desirable. What the systems engineer 
needs is attention and willingness to change or follow a new set of 
instructions. Few systems do not affect humans. A certain amount 
of the systems engineer's time will be expended in making the 
necessary adjustments (trade-offs) to satisfy the people directly con- 
cerned in the systems design. 

The final phase of the system study is that which extends the pilot 
operation, step by step, until it covers the full, operational scope it 
is intended to have. At each step of the way, the new system will have 
to be monitored because enlargement is in itself a test of adequacy. 
And finally, following complete implementation, there will be a 
policing period to insure that the system elements are functioning 
in use as they were planned. Policing, in this sense, makes the sys- 
tems analyst a control element — he acts as a system monitor. In 
systems implementation, policing is mainly concerned with the 
adaptation of individuals who may have to relearn their new rou- 
tines many times. This is a task requiring diplomacy as well as firm- 
ness. The systems engineer may police the system on the level of 
the individual operator, but more frequently works with super- 
vision in order not to infringe on supervisory prerogatives. 

The systems approach requires that there should be a plan for 
conducting a system study which borrows liberally from the scien- 
tific method. In addition, the systems approach dictates that a prob- 
lem be analyzed functionally and operationally and that it consider 
suitable alternatives. To achieve a high order of effectiveness in 
problem solving, the systems engineer attacks the task using all 
available tools. He looks at problems as potential systems requiring 
input, processing, output, control, and feedback. Among the first 
tasks is the isolation of the problem area by specifying the bound- 
aries around it. The systems approach dictates that a problem be 
attacked in an orderly way — first by investigation, then by estab- 
lishing a reliable hypothetical model of the problem. 



3 



Fundamentals of 
system design 



Investigation of a business system means the collection of volume 
and activity figures for each step of data processing and system de- 
sign. If these figures were being collected for a computer applica- 
tion, they would include the number of characters per record, the 
number of records in the file and their extension to determine the 
total size of the file in characters, separating alphabetic from nu- 
meric. In addition to this, frequency of processing would be stated. 
Data specifications would be included by obtaining samples of 
input, output, and file forms, perhaps utilizing a common detail 
specification sheet to bring all data on one subsystem together in 
one place. The natural accompaniment to this data would be flow 
charts describing the information transmission processes and data 
on existing semi-automatic or automatic data processing tech- 
niques. Where does the systems analyst go from here? 

The system design invariably reflects the philosophy of manage- 
ment. System design is a result of Management's policy, its knowl- 
edge of and orientation to company problems, and the excellence of 
the systems staff doing the work. Some philosophies of system de- 
sign are better than others under special circumstances. Since all 
firms cannot order large scale computers, a philosophy gauged to 
maximize the effectiveness of changing over to computers only 
would not be very useful. On the other hand, a philosophy dedicated 
to the status quo will never put the most sophisticated tools of the 
modern world at the service of industry. Some appreciation of use- 
ful goals in system design would clearly be of assistance. Figure 3-1 
illustrates some possible alternatives in systems design. 

41 



42 Fundamentals of system design 

ALTERNATIVES IN SYSTEM DESIGN: 
ONE-FOR-ONE SYSTEM CHANGEOVER 

Under this philosophy, the system analyst replaces the existing 
processor with some other processor justified either on the basis of 
faster reporting, more accurate reporting, or some similar improve- 
ment; savings in cost of data processing may or may not exist. This 
is a fairly common philosophy, although it is frequently disguised 
with admirable rationalization. The shortcomings of this approach 
are many. First, of course, is the failure to exploit available analytic 
techniques in the selection of the data processing system. The one- 
for-one philosophy substitutes two things in place of the analytic 
approach : 

(1) Reduction of elapsed time to install the new system. Example: 
Studies leading to the evaluation and selection of alternate 
equipment are eliminated. Whatever equipment advantages 
may exist, of one system versus another, are not explored. 
Equipment is thus selected on one of the following bases: 

a) Lowest in cost 

b) Outstanding (but unevaluated) performance 

c) Personal preference 

d) Integration with existing equipment (or enlargement of 
existing complement) 

(2) Minimum study costs to implement the changeover to the new 
system (see Figure 3-1). Example: Costs of studies to analyze 
equipment characteristics are eliminated. Costs to study an 
existing versus a proposed system are, by and large, bypassed. 



Continue Present Method 




— — — ^ 

One-for-One Changeover 




Design a New System 
Propose a new method of 
collecting and integrating 
data. 




New methods cost more 
than benefits derived. 


Changeover of present 
system to proposed sys- 
tem. Use same system 
but with faster processing. 










-«= 








Utilize Punched Cards 




Make a Computer Application 




Mani 
repl 


aal methods and costs 
iced in proposed system. 


Fully integrated systems 
have maxium flexibility 
and expandability. 



Fig. 3-1. Alternatives in designing the system. 



Fundamentals of system design 43 

Whatever systems problems are in existence, remain unsolved. 
Whatever cost advantages may be possible by and large are un- 
realized. 

Without exception, the one-for-one changeover results in far from 
optimal system design and/or equipment utilization; the systems 
approach, on the other hand, emphasizes that the system require- 
ments dictate the physical feature of the equipment to be utilized. 
An "off the shelf" philosophy has some further disadvantages. The 
time saved installing new systems and the consequent "savings" 
could be easily swept away by the need for system redesign, which in 
turn increases the study costs. Frequently, the one-for-one change- 
over fails because the increased speed of data processing cannot 
overcome the significantly higher costs of using more advanced data 
processing equipment. Sometimes this philosophy results in the 
selection of equipment that will be poorly utilized. The worst sit- 
uation that frequently results from this "minimum science" ap- 
proach is that the equipment complement is not well suited to the 
task. This does not say that the equipment utilized will not execute 
the task; the task, of course, will be executed — but awkwardly, in- 
efficiently, slowly, and probably at a higher per unit cost of infor- 
mation than some other complement of equipment. 

There are some cases where the one-for-one changeover will 
work: 

(1) When the input or output is being revised to accommodate sub- 
system changes and the function of the input or output proves 
to have a limited use or impact on the system as a whole. 

(2) When the changeover affects only the model or type of equip- 
ment employed, and both pieces of equipment operate in a 
similar or identical manner requiring no revisions elsewhere 
in the system. 

DESIGNING A NEW SYSTEM 

Diametrically opposite is the philosophy which looks upon any 
change as an opportunity to re-evaluate the soundness of an existing 
system. This may or may not result in the decision to expend funds 
for redesign. If the existing system is manual and existing volumes 
of data processing warrant investigation of equipment, almost in- 



44 Fundamentals of system design 

variably it will be necessary to study the system. One of the neces- 
sary considerations while selecting equipment is to consider not 
only current data processing loads but future, expanded loads as 
well. Thus, the redesign of a system provides an opportunity to 
consider more variables than would be ordinarily possible. Present 
and future data processing loads must, of course, be expressed quan- 
titatively. These quantities must be reconciled, to achieve the 
requisite goals, with the requirements of a minimum cost system. 
This rationale points to the need for optimizing a series of depend- 
ent (and sometimes, independent) variables. The variety of equip- 
ment currently available indicates a recognition of more special 
purpose needs than five years ago. The future will bring more 
diverse and modular equipment to the market — even less general in 
purpose than some existing equipment. Each complement of equip- 
ment is characterized by physical characteristics which are well 
known to the manufacturers — but which must also be made well 
known to the purchasers. The process of determining alternative 
equipment complement advantages is known as equipment evalua- 
tion (see Chapter 7). Familiarity with equipment characteristics 
demands special training. This must then be utilized objectively in 
the light of data processing systems requirements, since the system 
requirements must always dictate the processor. 

This is an imposing demand to place on personnel who carry the 
responsibility for systems design. As a result, some firms employ 
consultants specializing in data processing. The alternative route, 
less costly but replete with problems, is to select equipment based 
upon a set of rules — some of which may be quantifiable — almost 
without regard for the systems requirements. The only means by 
which systems requirements become known is the time consuming, 
costly, and orderly investigation and hypothesis of business facts. If 
the requisite skills do not exist within the organization, and if the 
move toward mechanization involves substantial outlays of capital, 
help from data processing specialists may prove to be the most in- 
expensive course of action. 

If the problem is to convert to a computer, the issue becomes 
critical. Computers are still so new that the vast majority of poten- 
tial candidates have little or no preparation for evaluating the use- 
fulness of one system over another. The use of a computer, except in 



Fundamentals of system design 45 

those special cases where an extremely sophisticated punched card 
(tabulating) installation has been in operation, will necessarily 
point a company toward extensive systems overhaul. For some com- 
panies, the most significant value in their impending computer 
program is the opportunity to reorganize departmental operations 
on a more modern, more economical basis. More and more experts 
in the data processing field agree that significant improvements are 
mainly in the area of systems design, under a philosophy that rec- 
ognizes the attributes of individual units of equipment. 

Benchmark problems 

When one or two applications dominate the total data processing 
requirement of an establishment, the benchmark approach is to 
order the equipment complement based upon the system require- 
ments of these few applications. The key word in this rationale is 
dominate. Ordinarily, it would not be realistic to order equipment 
unless sixty per cent (or more) of the total known data processing 
work load were known and analyzed. It follows that the benchmark 
approach would be perfectly adequate if the systems planned for 
conversion were sixty per cent or more. 

It is possible that the size of the problem would restrict the choice 
of equipment, or perhaps limit the choice to one complement of 
equipment. Should this be the case, the benchmark approach would 
eliminate other alternatives. When these special circumstances do 
not exist, however, the usefulness of the benchmark approach is 
severely restricted. Benchmark problems occur relatively seldom; 
thus, it is a highly specialized data processing problem. 

Mechanization in system design 

The move away from manual systems toward man-machine sys- 
tems is a step toward mechanization. Both utility and economy 
demand that such a move be considered in the light of the full sys- 
tem requirements. The insular view considers only one problem 
or a limited number of problems, without looking at either the 
future requirements or the full system requirements. Utility be- 
comes a factor because the limited use of equipment on a few 
tasks will leave it idle for long periods, which, in turn, effectively 



46 Fundamentals of system design 

raises the cost per unit of information. Economy becomes a factor 
because the uncoordinated acquisition of equipment may lead to 
parallel data processing facilities which can be marginally justified. 
Parallel installations are somewhat more difficult to centralize after 
equipment is operating. 

Figure 3-1 illustrates three possible alternatives {A, B, and C) in 
system design. These conclusions are a posteriori, i.e., after the fact. 
They are conclusions which grow out of alternatives 1 and 2. They 
are not the only possible viewpoints, since it follows that each 
business (system) will have its highly individual requirements. A 
rationale must be developed to guide the changeover from the 
original system to a proposed system, and this rationale must be 
supportable quantitatively. The move toward mechanization or 
automation need not take place in a specific, serial order, i.e., it is 
not necessary to move from all manual to all punched cards and 
then to a computer. Each individual case will have its own optimum 
solution. However, in any case, the changeover from one data proc- 
essing basis to another should be seen as raising the general level of 
data processing. This does not necessarily imply that the use of 
equipment is an all or none situation. 

The move toward mechanization should be made on a schedule. 
Individual applications with the highest payoff should be scheduled 
first, with other applications following as each area is converted. It 
is obvious that tabulating equipment and punched cards will never 
wholly replace humans in data processing. Thus, some combina- 
tion of A and B or C (Figure 3-1) will always remain even in the 
most advanced business systems. 

Raising the general level of data processing requires a careful 
look at all business activities and a judgment as to which parts of the 
business can or cannot benefit from mechanization. The intent of 
such a survey of data processing is to build in the following: 

(1) The potential for optimizing economy and utilization 

(2) The assurance of equal and full evaluation of all data process- 
ing tasks 

(3) The advantages which may result from organizational adjust- 
ments 

(4) The advantages which result from implementing the alternate 
which has the highest probability of success 



Fundamentals of system design 47 

This survey is what is commonly called a feasibility study (see Chap- 
ters 5 and 6), and can provide the long term plan of attack so neces- 
sary in mechanization of data processing. 

Both punched card and computer equipment offer something 
special to the systems engineer. They are, in many ways, ideal tools, 
since their physical design calls for an organization of data processing 
along the lines of the systems module. In a punched card application 
such as key-punching, a transaction of some kind will act as input. 
The processor will be a man-machine combination. The output 
will be the punched card. In the instance of a computer, a deck of 
punched cards or a magnetic tape carrying the same data may be 
introduced as input. The processor will be the computer and its 
operators. The output will be an updated file or files and the origi- 
nal input. The use of mechanization in data processing enlarges 
many times the problem solving ability of the business world. Both 
the size and complexity of normally difficult problems become less 
critical when equipment is at hand to aid in problem solving. 

The computer as processor 

The computer offers some unique advantages. For speed and 
reliability, manual systems will never be its equal. Interestingly 
enough, however, it is through the computer that the human being 
is able to attain the position in the business organization where his 
real capabilities can be properly utilized. Chapter 6 traces the de- 
velopment of this equipment. 

It is by now popularly recognized that computers make only the 
simplest of man-instructed decisions. The computer is the dumb 
beast which can do everything for which it has been especially de- 
signed — and can do it in a superior way. The individual must still 
be reckoned with as the ultimate decision maker. He can accept or 
reject the computer outputs, putting to work his judgment, experi- 
ence, maturity and intuition. This is precisely what makes the sys- 
tems purchaser (the person or group who must buy the system) so 
important in the design of the system. Not only must he have confi- 
dence in the system outputs, but the organization of the process 
must be completely useful and meaningful for the intended pur- 
pose. The systems analyst must sell the system, not only to gain 



4S Fundamentals of system design 

acceptance of his idea but to assure its continued use after the in- 
stallation is complete. The systems purchaser must be made to recog- 
nize that the system,, though capable of revision, should not be 
installed until it meets the exacting requirements with which it 
must deal when it is in full operation. 

The computer presents certain unique problems, mainly in the 
area of technical know-how. The central data processing facility 
presents other problems precisely because it is a central data proc- 
essing facility. Data processing is no longer a function scattered 
through a number of different departments, and the scheduling of 
work loads can become a major issue, each business department 
placing a certain priority of importance on each automated task. 
Here again, the systems designer must optimize the conflicting needs 
of interested parties. Scheduling also imposes a restriction on users 
of computer systems. All needs cannot be met on demand, but must 
either wait for scheduled processing or an unforeseen break in com- 
puter center operations when it is practical to interrupt normal 
routine. Once on the machine, the speed of problem solving fre- 
quently overcomes the delay of waiting. 

Computer capabilities such as simultaneous interrogation of 
many externally stored records places special emphasis on system 
integration. In the typical manual or punched card system, one 
operation is performed at a time: then the second operation; the 
third, and so on. Automatic control and other features enable the 
computer to do the first, second, third, and many other operations 
sequentially without stopping, except to report errors. Further, 
the computer can do more than one task at a time, reiving on in- 
ternal control to be the traffic monitor. The segments of integrated 
systems seen in Figures 4-1 and 4—2 show how a variety of inputs are 
planned to update several basic files. The ability to bring these di- 
verse items of information into one processing pass is the goal of 
integration. The special attributes and capabilities of four elec- 
tronic data processing systems are described in Chapter 7. 

SIDE EFFECTS 

Operations research specialists tell us that it is impossible to 
maximize more than one variable at a time. This is why it is essen- 



Fundamentals of system design 49 

tial to appraise carefully those things which happen in seemingly 
unrelated portions of the system when a system change is recom- 
mended. If a controlled laboratory experiment is viewed as a series 
of intimately related processes according to the systems concept, it 
is easy to understand the problem of side effects. 

Assume that the first output of an experiment provides satis- 
factory output, and can be repeated over and over again with identi- 
cal results. However, the economics are unfavorable, and other 
materials (inputs) must be tested to determine, (1) whether the 
same product (output) can be achieved, and (2) whether it can be 
achieved at a lower cost. When the experiment is conducted, a dif- 
ferent product is obtained. It is lower in cost but does not behave 
with the stability required. A third experiment produces a third 
product which is usable for the intended purpose but prohibitive 
in cost. Perhaps the four hundredth experiment will produce the 
desired product and the cost balanced in an optimal fashion. The 
large number of experiments have been occasioned by the follow- 
ing: 

(1) The variety of input possibilities 

(2) The nature of the processor which cannot cope with all types of 
input with equal efficiency 

(3) The resultant variation in output 

(4) The need to optimize cost, product quality and performance 

(5) The absence of controls to monitor all processes with equal ef- 
fectiveness 

(6) Lack of a priori knowledge to cope with all the new functional 
relationships established in the on-going process 

In the design of the business system, side effects occur frequently. 
By altering input or output requirements, the system designer has 
introduced a new series of considerations. The redesign of a process 
imposes on the designer the need to test his model, first theoretically 
and at a later date in some practical way, to remove the hazards of 
side effects. The ability to recognize undesirable side effects is in 
part conditioned by the effectiveness with which boundaries have 
been drawn. They will be related also to the uniform coarseness or 
fineness with which the system has been designed. Side effects in a 
redesigned system create undesirable symptoms, most of which can 



50 Fundamentals of system design 

be traced to inadequate collection or assembly of information prior 
to implementation. There is always some likelihood, in the process 
of consolidation and elimination of files or the development of 
multi-purpose reports and multiple forms from a common record- 
ing source, that side effects will occur. They should be expected 
and dealt with in an orderly way. 

Frequently, an undesirable symptom will become chronic, and 
the demand for system redesign will be made. The following are 
generally undesirable methods of solving such a problem: 

(1) Try a countermeasure, a typically short range solution which 
temporarily relieves the symptom. 

(2) Write an instruction to clarify the process in its existing form; 
add more rules. 

(3) Amplify an existing instruction. 

(4) Increase the number of copies of a report or form. 

(5) Add more equipment or personnel. 

(6) Add another input or output. 

(7) Borrow a system. 

(8) Provide an alternate process. 

(9) Add more checking, editing, or other operations. 

(10) Require more signatures, processing, and so on. 

(11) Increase the frequency of processing and output preparation. 

OPTIMAL SYSTEM DESIGN 

The weapons system is a composite of a number of electro- 
mechanical, electronic, hydraulic, physical and non-physical sub- 
systems' The operational characteristics of the physical systems have 
been made completely compatible with interrelated subsystems and 
with the over-all system. In this way, the unit subsystem functions 
as an integral part of the end item and is said to be fully integrated. 
The complete design accommodation of one part of a system to 
another works against parts interchangeability and design stand- 
ardization. Each weapons system has a special mission toward which 
the design effort has specialized it, although it may inadvertently 
contain characteristics similar to those of other weapons systems. In 
details, however, similarities cease to exist, and the unique require- 
ments of each system must dominate. 



Fundamentals of system design 51 

Business systems are dominated by similar considerations, 
whether they are recognized and exploited or not. It is not enough 
to say that the goal of every business is to maximize profit. Profit is 
one of many outputs and is the measure of a multitude of interact- 
ing systems which have been combined by a processor called the 
factory (Chapter 1). There are other desirable goals, and few busi- 
nesses can survive on a "profit-only" view in the business world. 

Systems purchasers 

Clearly, the "mission" as we know it from the example of the 
weapons system, was dictated by a purchaser, perhaps the Air Force, 
Navy, or Army. The requirements of the system purchaser are stated 
in the request for bid, to which the competitive prime contractors 
will respond by preparing proposals according to the rules, specifica- 
tions, and boundaries which have been laid down. When a contract 
is awarded, one of the functions of the prime contractor, who may 
also be a system manager in individual cases, is to interpret the 
contractual requirements for the ^^contractors. Thus, the prime 
contractor becomes the system purchaser, one level below the gov- 
ernment purchaser. The diagram for the system just outlined re- 
sembles Figure 3-2. The diagram for the system purchaser is given 



GOVERNMENT REQUIREMENTS; 
COMPANY PROCEDURES, RULES, 
QUALIFICATIONS 



EVALUATION TEAM 
RULES; PROCEDURES, 
CONFORMANCE TO BID 
REQUIREMENTS 



REQUEST 
FOR BID 



FACTORY 
A 



CLARIFICATION 

AND CORRECTIONS; 

CHANGES TO CONFORM 



-PROPOSAL 



Fig. 3-2. 



GOVERNMENT 
AGENCY 



NEGATIVE 
' RESULTS " 



AWARD 



in Figure 3-3. Every system without exception will have a system 
purchaser. The function of the system purchaser is to dictate the 
objective, purpose, or goal for which the system exists. Hence, in the 



52 Fundamentals of system design 



CONTROL -<- 



INPUT ■ 



PROCESSOR 



FEEDBACK' 



SYSTEM 
PURCHASER 



■►OUTPUT 



Fig. 3-3. 

example above, the system purchaser is the responsible government 
contracting agency (see Figure 3-4). In the system which illustrates 
the relationship of the prime contractor to the sub-contractors, the 
module might look something like Figure 3-5. Here are two inde- 



GOVERNMENT REQUIREMENTS; 
COMPANY PROCEDURES, RULES, 
QUALIFICATIONS 



REQUEST, 
FOR BID 



RESPONSIBLE 
GOVERNMENT 
CONTRACTING 
AGENCY 



FACTORY 
A 



■► PROPOSAL 



CLARIFICATION AND 
CORRECTIONS;CHANGES 
TO CONFORM 

Fig. 3-4. 



pendent forces acting as system purchasers. System purchasers act as 
a special type of control mechanism. Note that they do not operate 
on the processor directly but on the output and the controls. This 
may appear to say that the system purchaser is not interested in the 
means but only the end. This of course is not true; the system pur- 
chaser will monitor the means by evaluating the intermediate out- 
puts of subsystems. Control is not effectively vested in the system 



Fundamentals of system design 55 



EVALUATION TEAM 
RULES; PROCEDURES, 
CONFORMANCE TO BID 
REQUIREMENTS 



RESPONSIBLE GOVERNMENT 
CONTRACTING AGENCY; 
PRIME CONTRACTOR 



PRIME CONTRACTOR, 
SUB CONTRACTOR 
PROPOSAL 



GOVERNMENT 
AGENCY 



NEGATIVE 
RESULTS 



SUBCONTRACT 
AWARDS 



Fig. 3- 



j. 



purchaser; its role is to provide the objective for which the system 
exists. 

Note also the relationship of the system purchaser to the control 
mechanism of the system. An operational relationship must exist 
whereby the system purchaser's objective is consistent with controls 
which determine how the processor will satisfy the system require- 
ments. An example of this relationship is in the administration of 
technical orders. An individual contract specification might incor- 
porate a thousand technical orders into the quality requirements of 
a product. The responsible government agency may see fit to 
revamp many of the existing technical orders (as they did in the case 
of the ballistic missile programs) to implement the special objec- 
tives of a given program. The objectives of the ballistic missile pro- 
gram were made consistent with the standard control requirements 
(enforceable technical orders) of the contract by the action of the 
system purchaser. 

The system purchaser is a valid concept in the business world, as 
well. Behind the corporate executives who set the policy of the firm, 
stand many system purchasers: 

(1) The board of directors: Their function as system purchaser may 
be to examine current policy, relate it to profit, the market, long 
term objectives, and so on, and make new policies which might 
impose new constraints on output and control mechanisms. 
They are internal to the system. 

(2) The stockholders: Their function is more distantly related to 
the system, but the desire to buy or sell the corporation stock 
reflects a willingness to accept or reject the current policies 



54 Fundamentals of system design 

which are in force. The desire to buy might be a reflection of 
the desire to alter the existing corporate policy by gaining a 
voice in management. They are generally external to the system. 
(3) The users of product: The most effective system purchasers are 
those who actually pay money to use product. The test for prod- 
uct success is the market's willingness to pay the price of the 
product, to reorder the product, and to maintain the design of 
their product so that they must continue to use an output of an- 
other system. Product users are external to the system. 

In the last of the three cases just mentioned, the product user (sys- 
tem purchaser) may be said to energize (trigger) the feedback loop 
which sets in motion the special input mechanisms. In this same 
case, an optimization process is taking place. The need for product 
on the part of the system purchaser and the need for profit as an out- 
put of the system in point, are meeting in a trade-off. The knowl- 
edge that a system purchaser is always somewhere in the background 
must play an important part in determining how a company will 
attempt to secure a market and produce a product at a profit. 

There are systems purchasers for subsystems as well as systems. 
Some subsystems can be related directly to the product user. For 
instance, if the major objective of a production control subsystem 
is maintenance of schedule (see page 16), this would appear to 
satisfy the system purchaser's requirement that his order be shipped 
and delivered on time. Other subsystems, may relate outputs to 
systems purchasers within the organization, mainly the people who 
direct sections, departments, groups of departments, plants, or di- 
visions. The concept of the system purchaser provides another test 
of the validity of system design. In addition, it relates the system to 
something more than an objective, as such. Now it is possible for us 
to pass on the validity of an objective, having in mind some person 
or some group which stipulates the need or desire for a particular 
system. 

The users of systems may have conflicting requirements which 
will require the development of an optimum solution to resolve the 
problem. The systems analyst is the processor through which this 
optimization will take place. His role is to reconcile the conflicting 
demands and systems requirements as he sees them; he will deter- 
mine what system processor will best suit the problem. 






Fundamentals of system design 55 
Criteria and measures of effectiveness 

Another element is operating on output in some subsystems and 
systems. Criteria or measures of effectiveness supply a qualitative 
(non-numerical) or quantitative yardstick to gauge the effectiveness 
of output in satisfying system requirements. Measures are conceived 
concurrent with the design of the system. They must, like the system 
itself, be put into operation to determine if they are in fact measur- 
ing that which they were intended to measure. It also must be deter- 
mined whether the measure in use is the most indicative or critical 
yardstick which is available to measure the effectiveness of system 
operation. Some systems or subsystems will have only one measure; 
others will have several. A number of easily operated measures will 
have the advantage of providing more than one way to look at sys- 
tems outputs. Criteria are seen operating in Figure 3-6. The infer- 



control -*- 



CRITERIA 



INPUT- 



PROCESSOR 



FEEDBACK 



OUTPUT 



Fig. 3-6. 



ence here, because of his position in the system module, is that the 
system purchaser dictates the criteria. This is not necessarily so. 
Most of the system requirements which are stated as boundaries, 
such as 99 per cent system reliability for a missile, are control 
mechanisms. The ability of a missile to meet the on-line test routines 
of a missile site, for which the missile would be generally prepared 
only in its final acceptance testing, might be considered a criteria of 
this system (see Figure 3-7). In this instance, the using command 
would be the systems purchaser. Their specification that the missile 
meet certain on-line requirements may be resolved far earlier than 



*>6 Fundamentals of system design 



it\ 



MISSILE, 

GROUND SUPPORT 1 

EQUIPMENT, V 

PERSONNEL, / 

SUPPLIES / 



CONTRACT 
SPECIFICATIONS, 
CONFIGURATION, 
ETC. 



FACILITIES, MAN -POWER 
AND OTHER CONTRACTOR 
CAPABILITIES 



CHANGES AND 
•IMPROVEMENTS ■ 
IN SYSTEM DESIGN 

Fig. 3-7. 



ABILITY TO MEET 
ON-LINE SITE TEST 
REQUIREMENTS 



-►MISSILE SYSTEM 



the actual letting of the contract and the early design activities 
which characterize the beginning of a project of this type. 

The need for criteria for a system is, like the need of a system pur- 
chaser, mainly an operational requirement. The system analyst per- 
ceives a way of measuring the effectiveness of output when the out- 
put is known and the design of the system has been resolved. The 
ability to postulate and test a new system (see Chapter 2 which deals 
with system boundaries) utilizes measures of effectiveness as one of 
the means of determining whether or not system outputs are reli- 
able, accurate, and so on, and meet the objectives which may be 
quantifiable. 

Criteria may be qualitative. In business systems, it is likely that 
there will be a mixture of both qualitative and quantitative criteria. 
An example of the development of criteria in a business system, will 
illustrate this idea. 



Inventory status as criteria for a scheduling system 

There are three classes of inventory; raw or purchased, in-process, 
and finished. Most manufacturers have all types. Each is related to 
different parts of the scheduling system. Status of raw materials will 
be closely connected to the lead time of the ordering system. Raw 
materials inventory will tend to increase under the following condi- 
tions: 



Fundamentals of system design 57 

(1) Purchase requisitions or purchases are generated early or ma- 
terials are received early 

(2) End item requirements are overstated in the purchase requisi- 
tion 

(3) Lots are released late to production 

(4) Engineering changes which make inventory obsolete are not in- 
corporated promptly 

(5) Requirements are not combined to achieve optimum order 
quantities 

Thus, if the level of units and dollars invested in raw materials in- 
ventory are monitored through a reliable system, at least five major 
parts of the scheduling system will have been tested. There are 
more than five areas which can be tested by raw material inventory. 
The number of tests possible are related to the type of inventory 
record which is maintained, and the accessibility to inventory data. 
Work-in-process inventory will tend to fluctuate under the follow- 
ing conditions: 

(1) The availability of raw materials 

(2) The product mix-in-process 

(3) The number of lots and size of lots in-process 

(4) Availability of manpower, machine time, and tooling 

(5) The number of lots waiting for delayed operations 

Here again the tests would be for the number of units and dollars 
invested. Doing this adequately requires many more inputs than 
the previous example. For exactly this reason, maintenance of work- 
in-process is very difficult. Until the advent of electronic data proc- 
essing techniques, medium to large companies have found the day- 
to-day file maintenance of work-in-process was very costly, time 
consuming, and, in some cases, impossible. The five factors cited 
above themselves require subsystems with some dependence on the 
status of inventory. Late delivery of raw materials (item 1) can have 
an effect like a chain reaction throughout the entire factory system, 
although only a limited number of side effects have been noted 
here. 

The level of the finished goods inventory will be a reflection of 
the following: 



58 Fundamentals of system design 

(1) Anticipation of customer demand 

(2) Anticipation of rejections in the manufacturing process 

(3) Scheduling of parts, subassemblies, and final assembly 

(4) Anticipation of delays in test, rework, down time, and so on 

(5) The number of units being packed and shipped to fill orders on 
hand 

Note that the measure of effectiveness, which in this case would be 
the number of units or dollars of finished inventory, might show an 
undesirable increase. None of these five subsystems may be responsi- 
ble. Perhaps the increase is due to the failure of Production Control 
to schedule end-item rework on finished goods which must be made 
to conform to new field specifications before being shipped. 

Here are some additional inventory measures which would be 
useful to consider in the design of an inventory system: 

(1) Dollar value of purchased and manufactured parts 

(2) Number of dollars invested in parts 

(3) Number of dollars invested in dead or slow moving inventory; 
number of items in dead or slow moving inventory 

(4) Overtime hours and dollars related to shipments 

(5) Subcontract costs or delays related to shipments 

(6) Number of items on back order 

Measures of effectiveness can be coarse or fine. They can measure 
trends such as those demonstrated above, or they can be made very 
precise. Both are useful for their own special purposes at the option 
of the system designer — based upon the systems requirements. 
Measures or criteria, as cited, are yardsticks because they provide a 
continuing index to the way in which the system is operating. They 
are a part of the practical testing mechanisms usable as soon as the 
postulated system is ready to be tested. They continue to be useful 
through all stages of system design and system operation. 

Can completely manual systems be optimal? The answer is yes. 
The alternatives of each system problem must be analyzed and 
solved in their own frame of reference. Solving a problem on a 
manual level does not mean that alternatives do not exist, or that 
a fully integrated system cannot be achieved. There will always be 
limitations placed upon speed and accuracy of processing, but econ- 



Fundamentals of system design 59 

omy of system operation is another matter. The systems require- 
ments will dictate the number and rates of pay of the processors, 
human beings in this case. Optimal business system design means 
design of the system taking full advantage of integration possibil- 
ities, excluding the unnecessary or nonproductive portion of the 
system. In the design of the integrated system, the systems analyst is 
seeking the optimum combination of data, redundance, and static 
which enables the message to come through. In this information 
theory frame of reference, the data are the information to be proc- 
essed; the redundance is its format, display, and mode of transmis- 
sion; static becomes the variances or unanticipated special cases that 
arise and demand individual attention. 

A fully integrated system contains the most economic combina- 
tion of men, machines, and data that can be devised to execute a 
business routine. This is achieved only by linking individual data 
processing tasks intimately, so that they cease to function of and 
for themselves, but exist as a part of a total functioning unit. In this 
format, the outputs of the most elemental subsystems become inputs 
to higher order systems. In turn, the outputs of these higher order 
systems, become inputs to still higher order systems. Each level of 
the system operates under one or more control mechanisms, utiliz- 
ing the results of output wherever possible as a feedback to condi- 
tion and modify input. 

The fully integrated system provides not only the basic, requisite, 
system elements, but recognizes at each system level the need to 
equate the output with the satisfaction of an objective. The objec- 
tive must, in turn, be a requirement of some system purchaser, 
directly or indirectly concerned with the operation of the system. 
Likewise, each subsystem must meet the quantitative or qualitative 
test of criteria or measures of effectiveness which function as yard- 
sticks to gauge the reliability, accuracy, or performance of the 
output. 



4 



Postulating data processing 
systems 



ABSTRACTION IN THE BUSINESS WORLD 

Symbols may be used to represent any value, idea, or object, pro- 
viding only that the symbol has been carefully defined. Note that 
symbols do not necessarily have to represent quantities, but may 
also represent qualities. This is important, because the ability to 
work with symbols means quicker, more facile manipulation than is 
possible by extensive description of a condition, over and over again. 
The use of symbols in analysis of problems is, in some ways, similar 
to shorthand because it precisely and quickly records otherwise 
lengthy expressions. Symbols are used to represent quantitative 
values. It is also possible to set up symbols whose quantitative values 
will temporarily be unknown, and for the present may only be ex- 
pressed as qualitative ideas. 

The system modules are symbols of qualitative ideas. When they 
are interlinked to describe an on-going process, they become a 
model or miniature of the process itself. Model building shows the 
cause and effect relationships that exist between inputs and out- 
puts. The processor acts as the place where the functional relation- 
ships are expressed in the form of arithmetic or logical operations. 
The control mechanisms may be implied; but, in actual systems 
analysis, the controls must either be found or, if nonexistent, cre- 
ated. This is also true of feedback mechanisms. They frequently are 
expressed; but, when they are only implied, they must be found or 
created. The feedback property is essential since the modification 
of input as the result of actual experience (output), is highly desir- 
able. To the extent that feedback may cause input modification, it 

60 



Postulating data processing systems 61 

operates like a control mechanism. Unlike the control which oper- 
ates over the processor as the result of predetermined knowledge 
about how the system should operate, feedback is an after-the-fact 
control. Other controls on input, e.g., editing (filtering) of raw data, 
machine verification, and so on are possible. Analysis of the sub- 
systems where editing and machine verification occur would reveal 
that the actual control mechanism is a part of the process. 

On page 62, Figure 4-1 shows another type of abstraction. This 
chart illustrates the way in which a segment of a data processing 
system operates. This is a shorthand way of describing a series of 
individual operations that may number into the hundreds. Al- 
though all of the details are not here, adequate information has 
been abstracted to make it possible to understand what this system 
is intended to accomplish. 

At the outset of model formulation, the only knowledge of the 
problem may be the methodology to be employed in solving it. This 
is not bad. The predisposition to provide solutions before adequate 
facts are gathered and tested under a set of assumptions is all too 
common in the business world. The prime rule is to let the systems 
requirements dictate the system design. This, of course, puts the 
burden on investigating the facts and carefully interpreting their 
relative importance. Primary evidence in problem investigation 
will be copies of existing documents which energize the system (in- 
puts) and intermediate or final reports (outputs) which contain the 
results of the completed process. 

In Chapter 6 on page 108, you will find a Data Specification Sheet 
which is a suitable form on which to gather all of the data concern- 
ing a document, its use, its design, necessity, and character count. 
Some or all of the spaces provided may be relevant; the size or 
number of form copies may be adequate, or may not. The process of 
collecting data becomes extremely precise when carried to the level 
of individual processes. For instance, it is essential to look in detail 
at the circumstances under which the system is triggered into ac- 
tion. 

If a required form does not exist, then the system designer must 
postulate its introduction into the conceptual model of the system. 
He is saying, in effect, that this system cannot work unless there 
is some orderly way in which it can be energized. It will be up to 



Postulating data processing systems 63 

the system designer to provide adequate controls over the introduc- 
tion of this new subsystem, so that it will function as anticipated in 
the system design. 

DESIGN OF FEEDBACK LOOPS 

The identification of the feedback loop will follow the same 
general pattern. Feedback, which in business systems is a device for 
modification of input, must be considered by looking at the way in 
which the output is used. It is conceivable that systems can function 
properly without limiting in any way the effectiveness of the indi- 
vidual subsystem operation, but this is a rare situation. This does 
not mean that, in the ideal system, the processor should be called 
upon to do the same job twice — once, the way in which the input 
dictates, and, a second time, the way in which the modified input 
dictates. 

Feedback loops are variable in character. Figure 5-4 on page 87, 
indicates how the feedback loop takes an intermediate output which 
is transmitted to a planner; the act of providing him with inter- 
mediate output which can be modified prior to its use elsewhere, is 
feedback at its best. But there are other types. In Figure 4-2, a 
number of feedbacks are itemized: 

(1) The invoice from Shipping which will be compared with the 
purchase order 

(2) The stock ticket from Stores which will modify the work-in- 
process inventory 

(3) The scrap ticket, resulting from a rejection, which will also 
modify inventory 

(4) The cancellation or order reinstatement which will modify re- 
quirements as they may be expressed in work-in-process, raw 
materials, or finished goods 

(5) The job card (direct labor time ticket) which will adjust up- 
wards the cost expended in work-in-process, as each operation 
is completed 

(6) Shipping orders which will relieve inventory (in this case the 
item is shipped as soon as it leaves the work-in-process stage, 
there being no finished goods stage), as items are consigned to 
customers 



64 Postulating data processing systems 



FEEDBACKS 



PLAN SHEET 

-INITIATED BY SALES 

ORDER 



DIVISIONAL 
SHIPPING 
ORDER 



RECEIVING 

RECORD 

SUB CONTRACT 




Fig. 4-2. Segment of an integrated data processing system using 
punched cards or computer as the processor. 



Postulating data processing systems 65 

(7) Receiving records which will likewise modify requirements and 
stock status and will be compared to quantities on order. 

Note that feedbacks as just described have all been generated as 
a result of the output of some other subsystem. From the system 
designer's standpoint, the problem is to be certain that feedback 
actions are sustained and become as important a part of the system 
as any other. A company, convinced of this need, might originate an 
Action Notice to supplement the feedback; its purpose would be to 
record in the form of a report the serial numbers of transactions and 
a coded reference to the type of action that was required in order 
to close an information loop. This seems like placing undue empha- 
sis on one part of the system module; one way in which it can be 
justified is by examining the consequences of improper implemen- 
tation of feedback. 

Other feedbacks may originate from outputs — but, nevertheless, 
be outside the system. For instance, an employee receiving a pay- 
check for four days of work in a week where he worked five will soon 
correct a system error; few systems can be designed with as much 
feedback as payroll. The use of reliably interpreted data having no 
connection with system operation but maintaining a steady relation- 
ship (correlation) with it may provide effective feedback in a corpo- 
rate planning system. At the top business level, the profit and loss 
statement or the sales figures for an operational period provide 
great opportunities to use the feedback principle. If profits are 
lower in this period than any other, the cause may be traceable to 
higher costs of materials; following this hypothesis, the material 
values can be inserted in the cost equations and sales projections of 
coming periods to determine the future probable profits from opera- 
tions. 

An example of the use of sales data as feedback might be the firm 
that projects its price on an anticipated retail volume that does not 
materialize. If the price of the product is based on an annual 
volume, the firm must wait for reorders to determine the rate of 
production. Reorders, of course, will depend upon the sale of initial 
deliveries; if they do not turn over at the retail level, feedback will 
illustrate the reduced level of sale by smaller than anticipated 
orders — or none at all. The manufacturer will then have to move 



66 Postulating data processing systems 

quickly to redesign styles and assist retail outlets in moving mer- 
chandise, or else his business life may be threatened. 

Competition creates a number of feedback opportunities. If a 
manufacturer receives his orders to introduce a new line at a semi- 
annual market, one of his indications of success would be how well 
he fares in terms of new orders, first, compared to his pre-market 
estimate; second, compared to the intelligence he may be able to get 
about the status of his competition. Perhaps a more meaningful 
example would be two sets of dining room furniture in the same 
price range and otherwise suited to the same use within the home. 
One sells at rate x; the second sells at rate 3%. Each feedback must be 
interpreted by each manufacturer in the light of his own sales pro- 
gram. If, however, the rates of sale are different when each dining 
set is competing with a third model, or if sales rates are strikingly 
different in other outlets in the same or different cities, the manu- 
facturer will have data on which to create a new or, perhaps, more 
effective strategy. Most important, the sales strategy is going to be 
extremely difficult to design unless the manufacturer has provided 
the feedback loop to give him the desired data as a part of his sales 
program. 

DESIGN OF CONTROL MECHANISMS 

As a result of using the feedback loop, there may be additional 
action or there may be none. Action will be determined by the con- 
trol mechanisms (rules or procedures) which have been established. 
If a clerk is posting a manually operated inventory file, it is easy to 
see that feedback is dependent upon explicit rules such as the fol- 
lowing: 

(1) Sort transactions according to stock number: first, additions to 
stock; subtractions from stock following. 

(2) Enter all transactions affecting each stock number and strike a 
new inventory balance. 

(3) Recheck operation 2 for accuracy. 

(4) After posting, compare the new balance with the last with- 
drawal, inventory in-transit, replenishment cycle, and inven- 
tory minimum: 



Postulating data processing systems 67 

a) If inventory in-transit or stock on hand is adequate for cur- 
rent withdrawal rate, go to the next transaction. 

b) If inventory rate, level, or in-transit will not satisfy current 
requirements, generate a stock replenishment notice indi- 
cating inventory status, stock description, and so on. 

This control mechanism is working if the operator follows all rules 
explicitly, especially rule 3. These mechanisms are in the process 
part of the systems module. They are integral to the operation of 
the system. If this process has no external control, it is possible with 
twice the number of rules to incorporate and perpetuate errors in 
the system with far-reaching effects, as outputs become inputs to 
other systems. Ideally then, some external control to the system 
would be desirable, if it could be achieved at a cost compatible with 
the need. Control could be achieved in this way: 

Have a different clerk go through operation 3 above, making 
inventory adjustments if errors are detected. However this 
raises two points: 

a) What assurance is there that she will detect errors, if they 
exist? 

b) What assurance is there that the second clerk will not make 
errors as she makes adjustments to the stock records? 

If this sounds like an argument against using humans wherever and 
whenever possible, it is of course only partly true. Humans do make 
errors; about 2 per cent is the rate in efficient operators. But the fact 
is humans are the requisite to the processor in most systems. The 
cost of machine equipment to do the work of humans must be less 
than the costs generated by the humans operating the system. Sys- 
tems must be fairly large in size, complex, or heavily burdened with 
transactions before they become candidates for advancing to some 
level of automation. Thus, if we are to be dependent upon humans 
in a system design, we must attain control not by duplicating origi- 
nal operations which may double the cost of that operation, but by 
some summary operation done by the same or a second clerk. Here 
is a suggested control which will achieve the desired result: 

(1) Separate stock numbers against which postings have been made 
into lots, no lot to be larger than thirty transactions. 

(2) Sum all stock numbers added to or removed from stock to strike 

a total. 



68 Postulating data processing systems 

(3) Separate inventory cards against which postings have been 
made, into lots, comparable to operation 1. 

(4) Using inventory cards, go through operation 2. 

(5) If the sums of stock numbers in 2 and 4 do not agree, check the 
following: 

a) the sums of transactions and cards for errors in addition; 

b) postings to inventory cards. 

This is not the ideal way to control the accuracy of a system; it is 
only one way; there are many others besides this nonsense sum 
method. Although there is always the chance of compensating 
errors, some control has been achieved at a fairly low cost. It has 
nothing to do with the way in which the original process was done, 
and is thus not subject to the same kind of error. An interesting 
exercise would be to create several such alternate control mecha- 
nisms for this operation. 

Control may be achieved, with a little less fuss, in a punched card 
or computer system. In Chapters 6 and 7 which deal with com- 
puters, the logical and automatic control features take on special 
importance. You can see that it is possible to present to the computer 
a large number of very detailed rules, at many stages of the opera- 
tion, in the form of coded instructions. At tremendous speed and 
99.99 per cent plus reliability, the computer will post and report 
any errors or inconsistencies in input through exception outputs. 
Sometimes exceptions will be automatically generated in the form 
of cards; other times, from the computer operator's console type- 
writer. When a problem is dependent for its successful completion 
on early computations, the computer may be programmed to stop 
if an error of some type occurs. 

Control over the process can be achieved in a computer system in 
many ways: 

(1) Integral with hardware 

a) Bit or parity checks 

b) Logical checks 

c) Duplicate operations 

(2) Integral with system (external to hardware) 

a) Comparison of values with value limits 

b) High speed number checks (check digits, control sums, and 
so on) 

c) Decision rules and logical rules 






Postulating data processing systems 69 

Interpretation of results will of course return the burden of con- 
trol to the human, where it must ultimately reside. It is possible that 
the computer would go through a complex routine without stop- 
ping, providing an apparently perfect output. Once results are 
printed, they must be edited and reviewed for consistency and 
reasonableness. 1 

A fundamental premise under which the systems approach is 
utilized in business is its ability to bring the techniques of other 
disciplines to bear on qualitative problems. It is proposed that these 
techniques can become a framework in which large or small scale 
systems can be viewed. In addition, it is proposed that objects not 
customarily analyzed as systems can be looked at in this new frame- 
work, whether the object is a city, a war problem, a traffic problem, 
or the operation of a manufacturing business. 

The first requirement during the investigation will be to estab- 
lish a model of the system under study. In every case, the ingredients 
of the system, its inputs, must be isolated and carefully enumerated. 
Then the goals or purposes for which the system is supposed to exist, 
which we call outputs, must likewise be isolated and carefully 
enumerated. The ways in which the inputs are manipulated to ob- 
tain the outputs are the operations we have confined to the proc- 
essor. It is the processor which combines the inputs so that they fulfil 
the functions expected of each. These functional relationships may 
or may not be arranged to optimize the outputs; in some cases, they 
merely perform the steps necessary to obtain the outputs. A diagram 
of this much of the system module would look like Figure 4-3. 

The foregoing examples describe the procedure for dealing with 
both qualitative and quantitative problems. Industrial problems 

1 Dr. H. I. Ansoff, in his paper "State of the Art in Making Plans— Some Comments 
on the 111 -Structured System," distinguishes between two types of conceptual models. 
The outcome-oriented method would start with the development of a model to simu- 
late the behavior of the part of the business under study. The modes of behavior which 
would result from the selection of different strategies would be determined, and eval- 
uated. The alternate method is to devise a model which compares the salient character- 
istics of the strategies which contribute to the formulation of the system. Since the 
characteristics contribute to the realization of system objectives with varying force 
when they are evaluated in their several possible modes, the one that gets the highest 
rating is chosen. In the latter method, the outcomes are computed and only char- 
acteristics which induce favorable outcomes are used. This method called process- 
oriented, relies on the assumption that the more force which can be developed in a 
strategy, the greater is the likelihood of its success. Both methods are useful, although 
outcome oriented solutions may be easier to deal with. 



70 Postulating data processing systems 



INGREDIENTS = INPUTS • 



FUNCTIONAL 
RELATIONSHIPS 



OUTPUTS = GOALS 



Fig. 4-3. 

are generally qualitative and complex, and thus present a chore to 
the systems analyst. The systems model proceeds from top flow 
diagrams through several stages until it reaches the machine instruc- 
tion flow chart stage (see Chapter 5) and is in actual operation. In 
addition to flow charts, the model may be supported by mathe- 
matical or other symbolic data, which in themselves become models 
of some part of the total system. 

Only recently has it become possible to think of optimizing in- 
dustrial systems. This does not mean it has been successfully 
achieved on an organization-wide basis. But some industrial opera- 
tions lend themselves to mathematical-statistical models which 
have, as a part of their usefulness, the end goal of attaining an opti- 
mum combination of the following: 

(1) Fixed assets (plant and equipment) to end products 

(2) Inventory (of all types) to end products 

(3) Employment to manufacturing schedule 

(4) Financial requirements to short range and long range planning 

There are many possible optimums that may be stated. In this text, 
however, we are directing ourselves at the analytical tools by which 
it is possible to see processes as systems. Selection of factors to be 
optimized will be the job of the operations analyst (see Chapter 9). 



5 



Preparing for the 
systems study 



THE ASSIGNMENT 

The very first steps in the systems study can have an important im- 
pact on its success. The experienced analyst looks hard at two things: 

(1) Is there adequate time to prepare for tJie study? The time re- 
quirement can vary from a few hours to several weeks, and must 
be incorporated as a part of the over-all schedule, since complex 
tasks generally require more time than simple ones. 

(2) Is the assignment clear and precise? Understanding the assign- 
ment is the first order of business. Verbal assignments are most 
unreliable in interpretation. Written confirmation of a verbal 
assignment may sometimes prove that the same words can have 
more than one meaning to different people. Once the assign- 
ment is in written form, careful analysis of the premises and 
conclusions under which the assignment was made is essential. 
Examination of the written assignment may reveal that un- 
founded assumptions have been made. It is sometimes also true 
that, in giving assignments, the conclusion has already been 
built-in, and that, in effect, the assignment is to support an un- 
warranted or intuitive conclusion. 

In a department or section where a large number of investigations 
may be handled, it becomes necessary to set up a form to document 
individual tasks. Such a form is shown as Figure 5—1. This form 
provides a means of shuffling assignments and looking at the total 
work load in a variety of ways. It also provides a simple way of 
examining an individual problem from time to time to determine 
if the objectives, the means of achieving them, and the schedule for 
achieving them are still valid. 

71 



72 Preparing for the systems study 



WORK ASSIGNMENT 


Serial # 


Date Entered 


Requestor Date Due 


Department , Section or Group Requesting 


Name of Requestor 


~^\ Manual | [ E D P 


Task Title 


Task Order 


Authorized Hours 


Requestor Approval 


Acceptance Date 


Primary Responsibility 


Date Assigned 


Objective of Task 


Plan of Attack 


Step 


Manpower 


Type 


Hours 








Total 






Schedule of Accomplishment 



Fig. 5-1. Work Assignment Form. 



The time to conduct a study must be proportionate to its com- 
plexity; the problem may sometimes be so stated, during the process 
of determining an assignment, that it becomes clear it cannot be 
solved within the time available. The need to scale the problem to 
the time or the time to the problem becomes a factor in successful 



Preparing for the systems study 73 

analysis, if the engineer does not have adequate opportunity to do 
research. 

The Work Assignment Form suggests a way of examining the 
problem. It will clarify how much time to allow — or how to define 
the assignment so it can be done within requisite time constraints. 
The systems concept outlines six broad steps (Chapter 2) in the 
typical problem solving approach. Although all problems may not 
be adaptable to each step, the outline indicates a way of getting into 
the problem area. From this outline and from some knowledge of 
a problem, it is possible to begin testing the declared objective. A 
second step might be to assemble the criteria and list the assump- 
tions by which the objectives may be tested. A third step would be 
to determine if the requirements of the systems purchaser will be 
met by the assumptions implicit in the statement of the objective. 

Once these elements have been isolated and stated, it will then be 
possible to begin an outline of the way in which the study will be 
conducted. A so-called plan of attack will state not only how to go 
about the analysis operationally, but will state the minimum num- 
ber of activities that will require investigation in order to analyze 
the problem. This outline of the problem is a preliminary statement 
of the boundaries under which the problem will be studied. 

The method suggested for appraising the assignment can be de- 
veloped to a great degree of precision when it is the input to a pre- 
liminary proposal or estimate. The following activities are typical of 
an engineering department in the early phases of describing a re- 
search project: They must identify the skills, time, and costs to com- 
plete the task — and provide a schedule of completion. The systems 
analyst likewise must look on his assignment in an objective way. 
The systems analyst might suggest a review of the assignment, 
irrespective of the talent he brings to the task, if he cannot see the 
assignment output as the problem which should be under investiga- 
tion. 

Scaling the assignment to the time or manpower availability de- 
termines, in part, the coarseness or fineness with which a problem 
will be studied. It is conceivable that very complex problems can 
be subjected to very coarse analysis with acceptable results. This 
would only be true, however, if the problem was stated coarsely, and 
the solution supplied was uniformly coarse. Uniformity of depth in 



74 Preparing for the systems study 

analysis is critical and must be related to the needs of the problem. 
The statement of the assignment must be determined by the in- 
trinsic complexity of the problem carefully related to the depth of 
analysis required to supply an adequate solution. 

A simple simulation of the problem may be utilized to test the 
desirability of the chosen level of coarseness. The analyst may try 
the problem under the assumptions and criteria which support the 
objective of the task. Loose and undefined though the system may 
be in its assignment stages, such tests are desirable. In the absence 
of a completely conceived assignment, a half-conceived assignment 
with appropriate criteria and assumptions is clearly more desirable. 

Reviewing an assignment with other staff members as a part of 
the procedure of accepting tasks is also desirable. If the task to be 
analyzed can be logically explained to someone else without detec- 
tion of major unexplored or unanticipated problem areas, there is 
a good likelihood that the assignment has been adequately con- 
ceived. It may be desirable as a part of accepting an assignment, to 
present the problem as an assignment to a group of personnel with 
an objective orientation, under more or less formal circumstances. 
Making a formal statement out of a problem, and supporting the 
plan of attack as it is conceived under a given set of circumstances 
may bring to light any buried, unresolved areas of conflict. 

DEFINING THE PROBLEM AREA AND BOUNDARIES 

Adequate analysis of the problem will provide a careful statement 
of objectives and a plan of attack. The objectives must be closely re- 
lated to the assignment, whereas the plan of attack is integral with 
the problem area to be explored. In this way we are establishing the 
rules under which the analysis will proceed. The rules may be 
loosely defined as the boundaries beyond which it is not necessary 
or desirable to advance in order to solve the problem. 

There is a certain cost to be associated with every problem analy- 
sis. The cost boundaries should define how much money, time, and 
manpower can be allocated to a problem. The schedule of ac- 
complishment of a systems study is the time dimension under which 
the goals will be achieved. The budget is the dollar dimension; the 
staff is the manpower dimension. In a project of large scope the costs 



Preparing for the systems study 75 

can be stated as minimums and maximums. If the problem area can- 
not be adequately denned, a statement of the most optimistic or the 
most pessimistic complex of facilities to obtain a solution may be 
adequate. Cost-plus-fixed-fee contracts are in many respects like this, 
especially in research and development. Since the contracting gov- 
ernment agency does not expect a precise statement of anticipated 
costs, renegotiation is accepted by both parties as the by-product of 
contract termination. 

The statement of costs in terms of a set of flexible boundaries is 
not necessarily undesirable. Since changes in direction while a proj- 
ect is in process always entail some loss, an adequate statement of 
boundaries at the earliest possible time is necessary. Changes, when 
they occur, must be incorporated rapidly and efficiently, in light of 
objectives as they may prevail or as they may be modified. 

The boundary conditions as they are known in the G. W. Temp- 
lar and Company case (Chapter 10) are as follows: 

(1) Farber was an administrative assistant, probably a member of 
the corporate staff. 

(2) This task was to be executed for corporate level appraisal. 

(3) He was, for Mr. Templar's purposes, considered capable of 
handling the assignment. 

(4) Production Control, in the eyes of Templar or someone else in 
Management, required improvement. 

(5) Farber had three weeks in which to execute the assignment by 
himself. 

With these limitations, it would be desirable next to amplify on 
the various aspects of the problem in order to determine something 
about its size. With this need to define the location of boundaries 
around the problem, Farber might have decided the following: 

(1) What areas were dependent on Production Control. 

(2) Some maximum number of areas with which he would deal in 
his investigation. 

(3) The scope of study in each of the selected areas. 

(4) The priority of studies. 

(5) No area will be studied unless it contains over ten people. 



76 Preparing for the systems study 

(6) No area will be studied unless it is a prime input to Production 
Control. 

(7) Production Control is a strictly data processing function; its 
operation depends entirely on a variety of inputs. 

(8) Time to be spent in each area should be proportionate to the 
part of their budget which is related to some aspect of Produc- 
tion Control. 

You may not agree to some or all of Farber's boundaries. However, 
as an exercise, they are quite useful because they reflect the use of 
an objective approach. Here are some criteria or measures of effec- 
tiveness Farber might have conceived as he began to define the 
boundaries of his problem: 

(1) The time requirements of G. W. Templar and Company are 
met (are not met) by the present production control system. 

(2) The accuracy and reliability requirements of G. W. Templar 
and Company are met (are not met) by the present system. 

(3) Delays or errors related to the cost of the system versus the 
added costs to reduce either delays or errors. 

(4) How cheaply does the existing production control system pro- 
duce its outputs in terms of 

a) manpower outside Production Control 

b) manpower within Production Control 

c) Numbers of specially trained personnel; backup and train- 
ing required 

d) Flexibility of system design for sharp increases or decreases 
in volume of data processing. 

It is now clearer that the time available to analyze this problem will 
be a prime determinant in deciding how 7 far to go in getting answers. 
And yet, Management doesn't want a poorly integrated analysis, no 
matter how coarse it must be. It is conceivable that in three weeks 
Farber could only illustrate a few significant points to Mr. Tem- 
plar: 

(1) Production Control is extremely complex and is performing a 
large number of operations which appear to be of marginal 
value. 

(2) The cost of Production Control to Templar, according to some 
selected group of possible yardsticks, has risen (has decreased), 
and from this I draw the conclusion that further study should 
(should not) be made in these areas. 



Preparing for the systems study 77 

(3) The number of employees, number of forms used, number of 
reports produced, number of items controlled, time lags to gen- 
erate essential shop paperwork, and so on, behave in certain 
ways. Compared to past experience, where we achieved these 
results under other (less costly — more costly) conditions, the 
existing situation warrants (does not warrant) further study. 

Which point (there are more) shall he stress? Clearly we are doing 
some guessing in an attempt to understand all of the functional re- 
lationships which are represented by the processor, Production 
Control. Since the system compromises all of the inputs which con- 
tribute to it and forces them into a single mold, Faber must look to 
the purchaser of this system to make some judgment of the material 
to be presented after his three weeks of analysis. The purchaser in 
this case is Mr. Templar. However, other users of Production Con- 
trol outputs, whose performance may be based in part on how well 
they respond to Production Control outputs (schedules, for in- 
stance) might also be considered the purchasers of the system. 

A descriptive layout of the problem may be desirable. A form for 
this purpose is shown as Figure 5-2 on page 78. A check in the or- 
ganization column might mean that the form of the organization or 
its place in the total organization structure is to be examined. A 
check in the objectives column might be a question as to the validity 
of objectives or their consistency, clarity, or practicability in the 
light of certain corporate objectives. A check in the standards 
column might indicate the lack of adequate work measurement or 
the need for an overhaul of existing standards. 

A complete, descriptive layout may or may not be possible, de- 
pending on the investigator's specific experience in the problem 
area. In the initial phase of a study, where understanding and 
knowledge are certain to be at their lowest point, this task will be 
difficult, and yet it can be very helpful. The motivation for making 
the assignment must be expressed by someone in the organization, 
and this will provide some concept of the scope of the proposed 
study. 

Priority of effort and schedule 

In the simplest of problems, there is inevitably a choice, not only 
of objectives but of emphasis. Assuming that the boundary condi- 



78 Preparing for the systems study 





SCOPE OF STUDY 




Z 

O 
H 

< 

< 
o 

B 


CO 

> 

H 

O 

w 

•-3 

m 
o 


s 

s 

o 


CO 

a 

OS 

Q 
fcj 
O 

o 

Pi 


CO 

Q 

i 

< 


CO 

o 

H 
Z 

o 


H 

Ch 

H 
O 
H 

S3 
- 

z 


pq 

Q 


CO 

1 

to 
a 
o 
o 

E 


= 

O 


DEPARTMENTS (SECTIONS) 
FOR STUDY 


NO. 

EM- 
PLOYEES 


ANNUAL 
DOLLAR 
BUDGET 























































































































































































































































































































































































































































































Fig. 5-2. Analysis of scope of study. 

tions and objectives are well established, the next step is to create a 
step-by-step outline of how the ends will be achieved. Such an out- 
line need not be elaborate but should pick out the salient points, 
weighing them to provide ample time for analysis and report prepa- 
ration. 

One way to prepare such an outline would be to determine the 
organization units affected by the problem under consideration. 
Existing standard practices will provide a starting point. In the ab- 



j 



Preparing for the systems study 79 

sence of written procedures — and under any circumstances — the 
organization chart is a good place to begin. Sometimes, the analysis 
of functional responsibilities in the light of a specific problem may 
provide good leads on the study limits which must be set. 

The outline itself should be carried to the point where a well 
organized plan is conceived. This plan should indicate if possible 
the types of problems, as well as problem areas where investigation 
is anticipated. Each major problem or problem area should then be 
assigned a number of days in which to be carried out, using the three 
major categories of investigation, hypothesis, and implementation. 
The Work Assignment Form is usable for this purpose. Note that 
any problem or problem area may contain more than one task, 
while, in others, the number may be indeterminate and a flat allow- 
ance based on the best educated guess must be used. These numbers 
can be summed with a small safety factor added, where practical. 

The number of days required may be in excess of the time avail- 
able. The time requirement may be overestimated; if so, this can be 
corrected by reexamining estimates to bring them into line. Too 
fine an analysis may be planned — in terms of coarseness required — 
and again, some of the detail must be removed or the time estimates 
reduced. However, the time allowed by Management may under- 
estimate the problem scope; in which case they must accept a longer 
schedule of completion — or a coarser analysis not in keeping with 
the complexity of the problem. 

Scheduling a project is important because it creates motivation. 
The feeling of urgency in any project is essential because this has 
the effect of stimulating action from the investigator and attention 
from those being investigated. It is desirable to break up the proj- 
ect into milestones, so the schedule becomes meaningful in terms 
of the man-days allotted to each segment of the undertaking. It is 
never desirable to schedule only a start date and due date, unless 
the elapsed time between these two is very short or the number of 
things to be done within the scheduled period is very specific. Over- 
all schedules without intermediate due dates tend to be loose. Mile- 
stones act as control points and prevent a project from slowing 
down in one area — a slowdown in one area may jeopardize project 
completion on schedule. 



80 Preparing for the systems study 

METHOD OF OPERATION 

In the case study, G. W. Templar and Company, setting up the 
study could be a relatively simple matter. In other projects, how- 
ever, there may be a series of choices to make prior to completing 
the outline of the study. Consider a problem such as Marxson and 
Company (Chapter 12), where one of the study considerations is a 
choice of problems to work on first. In large scale studies, it is not 
uncommon to have to choose to work on some problems first and to 
put others in a secondary position. Adequate measures must be 
proposed, therefore, to describe which areas come first and why. 

The establishment of priorities may rest on the purely quanti- 
tative assessment of where the most dollars are being spent or where 
the probability of great savings are highest. A serial problem may 
also present itself. The choice may be based on the necessity for 
certain problem investigations to precede others because the solu- 
tion to problem Y is dependent upon the solution to problem X. 

Blocking out a problem may serve to align the subsystems in 
order of their interdependency (see Chapter 2, page 30). It is use- 
ful to remember the following while using the systems module in 
any large scale system: 

(1) There are many subsystems. 

(2) The integration of subsystems is essential to proper system op- 
eration. 

(3) The outputs of subsystems actually energize higher order, more 
complex systems. 

(4) Therefore, each system must be analyzed to expose the subsys- 
tems in their proper relationship to higher order systems, with 
a knowledge of the input-output requirements. 

Part of the master flow diagram which separates and locates the 
subsystems requiring analysis is shown in Figure 5-3 and is con- 
structed from the Wesley Engineering, Inc. case study (Chapter 18). 
In this case, one of the requirements is to look at the task of making- 
forecasts as a system. The analyst in examining the problem area 
wants to obtain a generalized picture of how the existing system 
operates. Each box represents a subsystem and a problem area. For 
instance, how is the headcount accumulated? How is it maintained? 



Preparing for the systems study 81 





1 1 

HEADCOUNT BY 
SECTION BY 
PROJECT 




HEADCOUNT BY 
DIVISION BY 
PROJECT 


































AVERAGE WAGE 
RATES BY JOB 
CLASSIFICATION 
AND PROJECT PER 
FORECAST PERIOD 




TOTAL AVAILABLE 
HOURS PER 
FORECAST PERIOD 




NET PRODUCTIVE 
HOURS PER 
FORECAST PERIOD 




























1 1 
PRESENT WAGE 
RATES BY JOB 
CLASSIFICATION 




ANTICIPATED 
WAGE RATES 
PER JOB 
CLASSIFICATION 















Fig. 5-3. 



How are changes introduced? How is it monitored? These questions 
and those which can properly be directed at any of the above sub- 
systems reflect the existence of or need for supporting subsystems. 
The arrow, coming into boxes marked 1, indicates that there are 
input requirements, some or all of which may depend on the pre- 
existence of other major systems. The headcount of a section must 
be predicated on these minimum factors: 

(1) Existing routine work load 

(2) Existing non-routine work load 

(3) Personnel safety factor (numbers of employees required to 
operate the system — allowing for vacation, sick leave, absentee- 
ism, and so on) 

(4) Personnel direction (supervisors, managers, project leaders, and 
coordinators, and so on) 

(5) A system for accumulating this data 

The ability to reduce personnel requirements to a format suitable 
for incorporation into a forecasting system, depends in part on the 
way in which the supporting subsystems have been organized. This 
interdependence works in the other direction as well. Higher order 
systems which will accept data must be designed so that their input 
requirements are compatible with lower order systems. 

The analyst preparing his plan of attack will be prepared to enter 
the problem area if, prior to actual investigation, he has blocked 
out his problem and looked at the problem areas objectively. The 
basic requirement in investigating data processing systems is to 



82 Preparing for the systems study 

make the problem manageable in terms of the assignment. The size 
of the problem will dictate the number of sub-problems to be ana- 
lyzed and hence have a profound effect on meeting the schedule. All 
of these conditions are, in turn, dependent upon an adequate state- 
ment of the problem and the boundaries under which the problem 
must be solved. 

New assignments are contingent on the status of existing assign- 
ments and current obligations. A schedule of obligations is valuable 
so that individual tasks can be rescheduled to retain high priority 
items at the top of the total list of assignments waiting. Since study 
assignments must be financed either by using departments or out of 
some general fund, scheduling of new tasks provides time to make 
funds available. 

Incoming assignments should be reviewed and accepted, subject 
to clarification of objectives or schedule, as necessary. The Work 
Assignment can be utilized as the standard recording medium, filed 
by subject, using department, type of study, or other means. Cross 
referencing is desirable, especially if many assignments are being 
worked upon or are ready for work at the same time. Work Assign- 
ment forms should be reviewed weekly for progress against the pre- 
determined schedule. 

SUGGESTED CHECKLISTS HELPFUL IN SETTING UP 
THE ASSIGNMENT 

I. Systems review checklist 

A. Purpose of Operations 

1. Have conditions changed since the operation was put into 
effect? 

2. Was the operation originally set up to correct a situation 
that has since been adjusted? 

3. Can we change the end result and eliminate the operation? 

4. Is the operation the result of habit? 

5. Is the cost of the operation justified by other factors? 

6. Is the operation created by an incomplete, previous or sub- 
sequent operation? 

7. Is the operation performed to satisfy the requirements of all 
or only a few of the persons in the system? 



Preparing for the systems study 83 

8. How necessary is the result accomplished by the operation? 

9. If it is a corrective operation, is it more costly than the diffi- 
culty it was designed to correct? 

10. How else can the result be secured? 

11. Are the results used as intended? 

12. Are all copies of forms or reports necessary? 

13. How many people or departments keep the same records? 

14. Do the report costs justify the results? 

15. Can the report be secured as the by-product of another op- 
eration? 

B. Machines and Equipment 

1. Does volume justify the purchase of general or special pur- 
pose equipment? 

2. Would savings effected over the average life of the equip- 
ment justify capital investment? 

3. Would other intangible factors, such as better customer serv- 
ice, valuable management reports, and so on, justify capital 
investment? 

4. Does the operation of the equipment require specialized per- 
sonnel or can existing personnel be retrained? 

5. Are existing machines operating close to capacity? What is 
per cent utilization? 

6. Is a central filing system indicated? 

7. Does existing equipment need repair? 

8. Does existing equipment have periodic maintenance and in- 
spection? Is it outdated? 

C. Data Processing 

1. What collating and sorting devices can be used to advan- 
tage? 

2. Would dictating equipment conserve a stenographer's time 
or eliminate bottlenecks? 

3. Would an automatic typewriter be more economical than 
the use of a manual typewriter? 

4. Is a duplicating process indicated? 

5. Is the volume of repetitive billing, statement, or payroll ad- 
dressing large enough to indicate pre-addressing from master 
addressing files? 

6. Can the addressing or duplicating equipment economically 
utilize automatic feeds and ejectors? 

7. Does the volume of mailing justify recommending a small 
or large postage meter? If a small manual one is used would 
volume justify a high-speed automatic postage machine? 



84 Preparing for the systems study 

8. Are typewriters in use suitable for the various specific jobs? 

9. Would the study suggest transferring any equipment to the 
other points where it can be used more effectively? 

10. Can office noise factors be reduced by soundproofing and the 
centralizing of high speed equipment in a separate room? 

11. Are vital records adequately protected against loss? 

12. Should permanent records be put on microfilm as fire pro- 
tection and /or to conserve filing space? 

13. Are circular, vertical, or horizontal filing devices the an- 
swer to some unusual reference or filing problems? 

14. What time-saving advantages would an intercommunication 
system effect? 

15. Would mechanical equipment be indicated to eliminate 
manual posting and recapping of columnar journals? 

16. Are desks, chairs, lighting, and so on, suitable for efficiency 
of the task performed? 

II. Infernal review checklist 

A. Organization 

1 . Is there an organization chart? 

2. Is the organization clear-cut and definite? 

3. Are there many or few layers of supervision? 

4. Is the functional write-up of organization units clear and 
complete? 

5. Are non-authorized functions being performed? 

6. Is over or under organization apparent? How? 

7. What numbers of people are reporting directly to each per- 
son on the chart? 

B. Procedures and Policy Manuals 

1. Have procedures been written and distributed? 

2. Are procedures complete and up-to-date? 

3. Are flow charts included? 

4. Are the procedures presented in a manner the worker under- 
stands? 

5. Do employees doing the tasks have copies? 

6. Do employees refer to procedures when problems arise? 

7. Are policy manuals maintained? 

C. Work Measurement and Production 

1 . What work measurement standards are used? 

2. Do the standards accurately reflect the work to be done? 

3. What overhead functions are not covered by standards? 



Preparing for the systems study 85 

4. Can standards be applied to these functions? 

5. Are the functions on a production basis? 

6. Can a production basis be applied to functions? 

7. Obtain production rates per man hour for previous periods. 

8. Obtain separate figures for overtime hours. 

9. Obtain overtime hours worked for these periods. 

10. What justification was furnished for overtime? 

11. Do the employees have production goals? 

12. Are employees aware of the goals? 

13. How many employees met the goals? 

14. How many employees passed the goals? By what amount? 

15. How many employees did not meet goals? By what amount? 

16. Is non-standard work budgeted? 

D. Schedules Backlog 

1. How is the workload scheduled and controlled? 

2. Is work introduced into system upon receipt? What are the 
factors which determine this? 

3. Determine date of oldest work in process. 

4. Are new batches completed before older batches? 

5. Obtain backlog figures for each of last four weeks. 

6. Are work priorities assigned? 

7. How are the priorities determined? 

E. Personnel 

1. What job classifications are used in the system under study? 

2. Obtain numbers of employees, by classification, now em- 
ployed. 

3. Obtain number of existing vacancies by classification. 

4. Is over or under staffing apparent? What yardstick points 
this up? 

5. Comment on employee morale, training, and supervision as 
they affect system operation. 

F. Facilities, Equipment, and Supplies 

1. Is the assigned area adequate? 

2. Is any part of the area not utilized? 

3. Does the layout lend itself to work flow? 

4. Can housekeeping and orderliness be improved? 

5. Are heating, lighting, ventilation, and so on adequate? 

6. Are the office furniture and equipment adequate? 

7. Is any equipment or furniture not being used? 

8. Is equipment being used improperly? 

9. What maintenance or servicing records are kept for equip- 
ment? 



86 Preparing for the systems study 

10. What was the cost of service and maintenance for the past 
three months? 

11. How are stocks of supplies and forms maintained? 

12. What controls are used for reordering or distribution? 

13. Is there apparent, excessive stocking? 

G. Records and Files 

1. Are files and records maintained in a satisfactory manner? 

2. Is the filing system adequate? 

3. Have record retention or destruction dates been established? 

4. Is there justification for retaining inactive or completed 
files? 

5. Can better utilization of inactive or completed file space be 
obtained? 

6. What records or files are microfilmed? 

7. Should any additional files or records be microfilmed? 

H. Budget and Cost 

1. What cost records are maintained? 

2. Are the records adequate? 

3. What cost controls exist? 

4. What is the basis for cost estimates? 

5. What safeguards control stamps and petty cash? 

6. Have costs been standardized? 

7. Are cost records used in budgeting? 

8. How do budget estimates compare with actual costs for past 
periods? 

I. General 

1. Obtain completed copies of all forms. 

2. Obtain completed copies of all reports required or origin- 
ated. 

3. Obtain copies or examples of rubber stamps in use. 



FLOW CHARTING 

Some special requisites exist if problem analysis leads to the use 
of electro-mechanical or electronic data processing equipment. 
Each assignment will have inherent characteristics that make it a 
candidate for manual or non-manual processing, although the 
choice of processor may have to wait until economic analysis proves 
the desirability of one over the other. In most cases, the first require- 
ment is the Master Flow Diagram, which is illustrated on page 62. 



Preparing for the systems study 8y 

This diagram broadly defines the subsystems as elements of the 
higher order system, from the beginning of the process to its end. 
Because this chart uses rectangular blocks to show subsystems, it 
serves only as a generalized model of the system. Its prime useful- 
ness is in providing the opportunity to look at the total problem 
early in the investigation period. 

The Top Process Flow Chart (see Figure 5-4) is the second step 



PLANNING DEPARTMENT 


INPUT 


PROCESS 


OUTPUT 




















EXAMINE 
RESULTS 


.@- 














KSPOHT 


















NEW 










REQUEST 






























|(!> 




INPUT 
FORM 




<£>- 


MACHINE 

LANGUAGE 

DATA 


— (T)- 




PROCESSOR 


<7\ 




/ MACHINE 
LANGUAGE 
DATA 






A 




\L> 






(7) 






( BASIC 
DATA 
FILES 






OUTPUT 
FORM 












^ 










(T) Request for evaluation of 
proposed change is sent to 
Planning Department. 


(3) Data on Input Form 
converted to machine 
language. 


(7) Calculations performed 
and data prepared for output. 


(V) Machine language 
data converted to printed 
format of Output Form. 


(2) Planner studies, reduces in- 


(4) Entered into proces- 


(?) Information extracted. 




formation to Input Form. 


sor. 


(5) Processor consults basic 




(9) Planner examines simulated data. 
May consider alternatives and repeat 




data files. Simulates the 
presence of new data In the 




simulation process. 








orij 


Planner repo 
■lnating agen 


rts fin 


lings t 




























Fig. 5-4. Top process flow chart. This shows how an inquiry is intro- 
duced to a simple system. 

in defining the problem under investigation. This is a gross infor- 
mation flow diagram, indicating the forms, processing steps, and 
outputs associated with a problem area. The top chart — the chart 
which describes the total process — must make ample reference to 
the Detail Process Flow Charts which contain complete data on the 
most minute subsystems which are under analysis. Detail flow charts 
and the top flow chart should form a package, together with notes, 
sample forms, reports, and other documentary evidence, to provide 
a full picture of the problem area. 

If the problem is resolved into one requiring the use of electro- 



88 Preparing for the systems study 

mechanical or electronic data processors, Top Process Flow Charts 
will be used as the basis for drawing a Top Computer Flow Chart, 
Detail Computer Flow Charts, and Detail Machine Operation Flow 
Charts. A set of suggested symbols for these flow charts is shown in 
Figure 5-5. A sample Detail Computer Flow Chart is shown as 
Figure 5-6. 



PROCESS FLOW CHART 



COMPUTER PROGRAM 
FLOW CHART 



MACHINE OPERATION FLOW CHART 



ORIGINATOR 



o 



DOCUMENT 




OPERATION 



OPERATING PROCEDURE 



DECISION 




EXCEPTION 



FILE: 
P = Permanent 
T = Temporary 
or Suspense 




CONNECTOR 
O 

DIRECTION OF FLOW 



CONTROL 



o 



CONNECTOR 

o 

DECISIONS, SWITCHES, 
BRANCHES 




DIRECTION OF FLOW 



OPERATIONS: 
FD= Fixed Decimal 
FP = Floating Point 



MODIFICATIONS 



O 



INPUT - OUTPUT 
Magnetic Tape 



LJ 



INPUT - OUTPUT 
Punched Card 



EXCEPTIONS 



( ) 



CLERICAL MAGNETIC TAPE 



( ) 

MAGNETIC DRUM 
TOR 



Q 



SORTER 



o 



DOCUMENT 

ORIGINATING 

MACHINE 



REPORT 




LINE PRINTER 



TAPE TO CARD 



KEY PUNCH 



D 



CARD TO TAPE 



TRANSCEIVER 




FILE CABINET 



P 



Fig. 5-5. Standard data processing symbols. 



, 



START Q/ - *" 



Input 



Correct Payroll 
Number ? 



NO 



YES 



Hours x Rate 
= Gross 



Print "Input" and 
Tape "Record" 

Numbers 



Last Check 



YES 



Halt 



NO 









YES 












Adjust uross 




1 


r NO 


1 




Link Back 


YES 


Previous Yearly 
Earnings;>$4200? 










YES 




1 


r NO 




Obtain Earnings 
to Date 




* 




Earnings to 
Date>$4200? 


Compute 
F.I.C.A. 




Ino 


1 


Link Back 


Compute 
F.I.C.A. 






1 




Compute 






Ta 


-X 









Compute 

Total 
Deductions 

zxz 



Compute 
Net Salary 



0^L 



Print Check 
and Stub 



Last Check ? 



YES 



Halt 



Fig. 5-6. Detail computer flow chart for payroll. 



90 Preparing for the systems study 

All flow charts should be prepared on standard format vellums 
carrying suitable reference data so they can be correlated. Ample 
cross reference should be provided to allow personnel to move freely 
from notes to flow charts, flow charts to sample forms or reports, and 
so on. In large scale studies, notes, flow charts, and sample materials 
are gathered in applications folders and paper bound for ready 
reference or modification. The organization of data for easy refer- 
ence requires the development of procedures and codes by which 
card formats, applications, keypunch instructions, and so on can 
be standardized. 

STAFFING THE PROJECT 

Today's technology demands the use of varied skills in problem 
solving. Seldom can all the requisite skills be found in one person. 
Frequently, several people will make substantial contributions to a 
system study. On page 164, a particular study requiring three dis- 
tinct skills is diagrammed. In Chapter 6, page 103, a typical data 
processing facility organization chart is shown. Taken together, 
these exhibits demonstrate how the mixed team is formed. Table 5- 
1 shows the steps required to implement a computer application 
after the postulated system has been designed. This demonstrates 
another organization of a mixed team: Here, technical skills have 
been provided from outside a department, but department mem- 
bers have actively participated in the project itself. In some areas the 
department members work alone; in others, the technical experts 
are, likewise, alone; in some work, their activities are combined. 

The fundamental need is, of course, to have access to personnel 
for specialized uses. Where technical skills are unavailable within 
an organization, it may become desirable to employ outsiders who 
can furnish such skills for a short term. To recognize the need for 
special skill requirements and to determine whether or not they are 
available are two distinct activities. Both are functions of the project 
leader; but, frequently, the men working on the problem bring 
them to his attention. 

The distribution of assignments to team members sets up the 
adjacent problem of maintaining a good fix on the selected goals. 
When the project leader is a working member of the team, super- 



Preparing for the systems study 91 



TABLE 5-1 

Basic Steps Required to Implement 

A Computer Application After 

Postulated System Design 



Departmental Activities 



5. Begin data collection. 



7. Place purchase orders 
for forms. 



12. Document procedures 
to provide permanent sys- 
tem capability. 



14. Monitor availability of 
new input data. 



Combined Departmental 
and Technical Activities 



2. Simulate detail system 
design to uncover prob- 
lems. 



4. Design final input 
and output forms. 



10. Check results of sys- 
tem operation as each 
segment is sucessfully 
programmed. 

1 1 . Complete system 
operation and develop- 
ment of machine oper- 
ating instructions. 



13. Proof finished out- 
puts. 



Technical Activities 



1. Determine detailed sy; 
tern design. 



3. Design punched card 
layouts; keypunch for- 



6. Block diagram, pro- 
gram, and code; prepare 
for machine operation. 



8. Keypunch inputs as 
available. 

9. Complete technical de- 
bugging and program op- 



vision and over-all project direction can become a burden. Day to 
day activities must be monitored and tested for conformity to ulti- 
mate objectives and schedules. Individuals tend to participate in 
mixed teams according to some a priori conception of their skill, 



92 Preparing for the systems study 

role in the group, and understanding of the problem. The project 
leader's task is to mold individuals into a cohesive team working 
toward well defined milestones. 

Selling the assignment 

Too frequently, the project leader does not lead. He accepts an 
assignment, works at it, and delivers a result — all of this in an ade- 
quate but uninspired way. The great majority of failures in systems 
studies are failures in human relations. Either the project leader is 
not sold on what he must do, or his team members have not been 
sold on the importance of their individual roles. Very frequently, 
the primary beneficiary of the system study is not sold on the bene- 
fits to be derived or the goals which have been set up. 

Inevitably, the failure to sell the study is reflected in the quality 
of the results and the effectiveness of the system in operation. Nota- 
ble failures occur in selling the selected course of action to Manage- 
ment, and countless, great ideas never see the light of day because 
of the failure to put them across. Since recommendation to expend 
funds frequently accompanies the presentation of a new program 
to Management, the need to sell the idea becomes doubly im- 
portant. Not only may personnel be displaced or reassigned with the 
introduction of a new system, but expenditure of dollars will occur. 
How then does one "sell" the assignment? 

The selling must begin at the lowest level. Confidence in the 
assignment is primary. This has something to do with the recog- 
nition of the need and the knowledge that, if the problem is solved 
and implemented, good will result. Not only willingness but ability 
to do the work is an important factor. Personnel trained and capable 
to undertake the task will find it easier to accept the problems which 
will arise while the study is in progress. 



6 



Electronic data processing 
systems 



The systems approach is best described as a way of thinking about 
problems. From the range of cases in the following chapters, it is 
clear that the systems approach is not confined to large scale prob- 
lems or problems of a purely quantitative nature. There is one class 
of problems, however, which can be attacked using systems thinking 
with high quality results. These problems are in the area of elec- 
tronic data processing. This section of the text deals with the use of 
electronic data processors in the systems design. 

EARLY DATA PROCESSORS 

Even though early equipment enhanced the performance of data 
processing tasks, there were still many drawbacks. The foremost 
of these was the inability to perform more than one function at a 
time. To overcome this disadvantage, later development empha- 
sized equipment that could perform more than one data processing 
function in a sequence of operations. The first pieces of equipment 
produced during this stage were adding and listing machines. These 
machines combined the functions of computing and recording. 
Another well-known piece of equipment that combined the func- 
tions of computing and recording is the cash register. The latter was 
developed to provide automatic records of retail transactions, to 
issue receipts to customers, and to give detailed information to aid 
Management. 

Another significant advance was the development of the ma- 
chines of the billing, bookkeeping, and accounting types. These ma- 
chines combined the functions of computing, sorting, distributing, 

93 



94 Electronic data processing systems 

and recording. They were designed to deal with such business prob- 
lems as preparation of invoices, reports, and business documents, 
and were useful in that they introduced more speed, neatness, and 
orderliness to accounting records. 

A disadvantage of the more advanced equipment was that the 
mechanical functions were not closely integrated. AVhen informa- 
tion moved from one machine which would perform one or more 
functions to another machine which would perform other func- 
tions, it was necessary to have a human being do the actual transcrip- 
tion of the information into the second machine. It was also neces- 
sary to have a human perform the actual transcriptions from the 
documents generated during each step of the data processing. This 
introduced a source of error and was a time consuming operation. 

The first attempt to overcome this lack of compatibility was 
punched card equipment. 1 The punched card introduced compati- 
bility to machines of the same kind. That is, all IBM punched card 
machines are compatible with each other, but not with all other 
types of equipment. The punched card provides a method of mov- 
ing from one machine which would perform one or more functions, 
to other machines which would perform other functions, without 
having a human do the actual transcription of the information 
punched into cards during each step, although it is necessary to have 
a human punch the information from the original source docu- 
ment into the original cards before the processing begins. Then the 
punched cards, both the originals and others generated by the proc- 
essing, provide a medium for the data processing without human 
transcription. 

Punched cards do not have unlimited compatibility. To over- 
come their shortcomings, a system of providing compatibility to 
a wide range of equipment through the medium of punched paper 
tape was developed. Punched paper tape carries the information 
from one machine which performs one or more functions, to other 
machines which perform other functions, thereby eliminating the 
need for human transcription during each step of the processing. 
This system makes it possible to use accounting machines, type- 
writers, cash registers, addressing equipment and calculators in com- 

1 Manufactured by International Business Machines Corporation and by Reming- 
ton-Rand, a division of Sperry-Rand Corporation. 



Electronic data processing systems 95 

bination with each other and with punched card equipment. The 
technique of using the punched paper tape offers possibilities of 
overcoming communications problems and providing a common 
language for existing types of mechanical equipment. 

ELECTRONIC DATA PROCESSORS 

Even with punched card equipment and integrated data process- 
ing, there was still a serious drawback. Human error had not been 
eliminated between data processing functions. With punched card 
equipment, it was necessary to have an operator transport cards 
from one machine to another, and to initiate subsequent opera- 
tions. This was also required when punched paper tape was being 
used. The goal was to eliminate human intervention between suc- 
cessive operations and to have a machine capable of performing 
many data processing functions. Although the automatic calculators 
proved highly successful, they had several disadvantages. One was 
that they were too slow. This slowness resulted from the fact that 
machines like the Mark III and Mark IV used mechanical counter 
wheels for arithmetic elements and for storage. The Bell machines 
used electro-mechanical relays for arithmetic elements and for stor- 
age. To overcome this slowness, electronic parts were used to replace 
the mechanical and electro-mechanical parts of automatic calcu- 
lators. 

The first development during this stage was the ENIAC. It was all 
electronic except for input, output, and certain switching functions, 
and had the features of automatic control and high speed. The 
ENIAC was the grandparent of all present-day electronic com- 
puters, and its success caused other electronic computer projects to 
be pushed to completion. In fact, it was originally designed for one 
specific type of problem, namely the computation of ballistic tables 
with no possible business use. ENIAC was an inflexible machine 
because it required a considerable amount of time and effort to 
change problems. The next step was to overcome these disadvan- 
tages and still maintain the features of high speed and automatic 
control. 

The next significant development came from Eckert, Mauchley, 
and Von Neumann. They conceived of the idea of incorporating the 



96 Electronic data processing systems 

instructions for the automatic control of the equipment in a digital 
memory device, together with the data to be processed. This indi- 
cated that existing equipment required much more storage capacity. 
Enlarged storage capacity greatly increased the variety and types of 
problems that could be handled. In addition, the incorporation of 
the instructions in the digital storage device also introduced a 
greater degree of flexibility to the equipment. With this concept, 
it was now possible to read the control instructions into the machine 
in the same manner in which the data was introduced. This facili- 
tated the changing of problems. 

Out of this advance came the design for the ED VAC, giving it the 
distinction of being the first piece of equipment to hold its own 
program of control instructions in its memory. However, the first 
stored-program computer to operate satisfactorily was the EDSAC 
at the Mathematical Laboratory of the University of Cambridge. 
Both of these machines incorporated the stored program feature 
and were the first electronic automatic digital computers of the 
general purpose variety. 

CHARACTERISTICS OF ELECTRONIC EQUIPMENT 

Electronic data processing equipment has the capability to do 
the following things: 

(1) Perform basic arithmetic operations of addition, subtraction, 
multiplication, and division. 

(2) Perform certain logical operations such as the selection of the 
larger of two numbers or the choosing of alternatives depend- 
ing on previous conditions. 

(3) Record, remember, and recall data in its storage facilities. 

(4) Communicate the results of its operations, in intelligible form 
or to other machines. 

(5) Accept information and instructions. 

(6) Direct itself in a predetermined manner, without human in- 
tervention, through its stored program. 

(7) Check the results of its operations and report when the checks 
are not functioning satisfactorily. 



Electronic data processing systems 97 

As a result of these abilities, electronic data processing equipment 
can simulate human beings doing many tasks. In fact, computers 
can do most of the clerical operations performed in manual and 
electro-mechanical data processing systems, including the making 
of routine decisions. These abilities make electronic data processing 
a potent tool. However, there are some things computers cannot do. 
These are at least as important to understand: 

(1) They can perform operations only if instructions on how and 
when to do them have been given. 

(2) They cannot construct their own sequence of control instruc- 
tions, though they can modify instructions. 

(3) Computers are not error free, although the frequency of errors 
per unit of work is much smaller than the rate of human errors, 
if a human were to undertake the identical job. 

(4) Computers do repetitive operations more efficiently than non- 
repetitive operations. Therefore, the time required to prepare 
a problem may make it uneconomical to undertake short term 
tasks. New programming techniques are working against this 
limitation. 

(5) Computers cannot sort efficiently. To sort a set of numbers into 
an ascending sequence requires many passes. If the set is large, 
this is obviously awkward and inefficient. It is still much faster 
than the human rate. New equipment has attacked this limita- 
tion with substantial improvements in sorting speed. 

Speed 

One of the inherent characteristics of electronic data processing 
equipment is its high speed. The high speed of electronic switching 
circuits makes possible a fantastic rate of data processing operations. 
Arithmetic operations on numbers or the comparison of two num- 
bers are performed at the rate of thousands per second. Computer 
operations can be performed at rates up to 400,000 times those of 
manual data processing. These rates apply to the internal opera- 
tions of electronic data processing equipment. 

The speed of the input-output operations is not as great as that of 
the internal operations. However, input-output speeds become very 
important because data must be introduced to the computer 
through some input device and sent from the central computer to 



98 Electronic data processing systems 

an output device. Thus, to some extent, they control the over-all 
system speed. 

To overcome this limitation of the over-all speed of the entire 
data processing system by the input-output equipment, several 
things have been done. One thing is to use magnetic tape as much 
as possible for large volume input-output operations. Since mag- 
netic tape is by far the fastest input-output medium, it offers a 
greater over-all system speed than could be obtained by other input- 
output mediums. The use of buffers between the input-output units 
and the internal processing units is now common. Buffers offer the 
advantage of simultaneous operation of the input-output units and 
the internal processing units. This also can contribute to the greater 
over-all speed of the data processing operation. 

Even though the input-output units can cause a slow down of the 
over-all data processing speed, it is still possible to maintain rates 
up to thousands of times those of manual or mechanical methods. 
High speed electronic data processing may make it possible to report 
business activities at less cost than manual or mechanical methods. 
It also makes possible the solution of business problems that have 
not been susceptible to manual methods in the past. 

Automatic operation 

Another characteristic of electronic data processing equipment 
is its ability to perform programmed, automatic, data processing. 
Once the equipment has been loaded with the information to be 
processed and told what to do, it can process this information with- 
out further human intervention. Automatic operation is the result 
of a set of elemental rules of operations that have been designed and 
wired into the equipment. The computer uses its logical circuits 
and its man-given information to do a predetermined sequence of 
operations. Automatic operation also contributes to the high speed 
of electronic data processing. The computer can go from one opera- 
tion to another at electronic speeds, since human intervention is not 
necessary between steps. 

Programming is the process of preparing the procedures for 
processing data. The programmer breaks down a complex opera- 
tion into a sequence of simple operations. He further isolates the 
sequence of procedures necessary to complete an operation. The 



Electronic data processing systems 99 

coder translates the data processing procedural scheme into a de- 
tailed list of instructions that the equipment can handle. 

The instruction code that is used in preparing the programs for 
solving the problems, determines the principle operating steps of 
the electronic data processing equipment. Each instruction is de- 
signed to accomplish an elementary data processing operation on 
one or more pieces of information selected from the storage unit, 
and /or to transfer the result of an operation to the storage unit. 

Each instruction generates a group of control signals. These con- 
trol signals select the information on which the operation is to be 
performed from their storage locations. In addition, the control 
signals can transfer the result of the operation to the appointed 
storage location, and they can select the next instruction from the 
storage unit. As a result of this, the instruction that represents the 
control signals has to specify the operation to be performed and the 
storage locations of the operands. It can also specify where the result 
is to be located within the storage unit and where the next instruc- 
tion is located. 

In machine language programming, one instruction or command 
generally equals one minute step in the process to be carried out. A 
new factor in the programming field is called English language pro- 
gramming. The major manufacturers of computers have developed 
new techniques where one instruction is equal to several — even 
many — detail instructions; in addition, these instructions can be 
written, more or less, in the English language. That is, the format of 
the instruction and its content can be understood, whereas in the 
past code letters or numbers were used. This is a major step in com- 
puter technology and has reduced the former, sizable costs associ- 
ated with programming. As newer techniques are developed in 
association with hardware which has been designed to accommodate 
the new programming techniques, the cost may be reduced still 
further. 



Flexibility 

Another characteristic of electronic data processing equipment is 
its flexibility. Computers are designed so they can operate on a 
variety of problems. Flexibility arises from the ability to change the 



100 Electronic data processing systems 

sequence of control instructions. By changing the sequence of con- 
trol instructions, it is possible to change the type of problem being 
solved. Electronic data processing equipment can perform any oper- 
ation that can be expressed in terms of sequence of control instruc- 
tions belonging to the instruction code of the equipment. However, 
operation has to be such that it fits within the limits set by the 
equipment's storage capacity. If the instruction code includes the 
arithmetic operations of addition, subtraction, multiplication, divi- 
sion, and a few simple logical operations, the equipment can per- 
form practically every data processing operation. 

A piece of equipment that has this data processing capability is 
called a general purpose machine. Most of the recent electronic 
data processors are general purpose machines. General purpose ma- 
chines can handle only a restricted class of problems efficiently. 
These limitations are a result of the equipment's instruction code, 
the capacity of its storage units, and the speed of its operations. 

There is another class of equipment serving useful purposes in 
electronic data processing. This equipment is called special-purpose 
and is usually designed and engineered for special applications. 
They are not flexible since they can only perform the more limited 
operations for which they were designed. The instruction code is 
built into the equipment and cannot be changed unless an engineer- 
ing change is made. ERMA, a computer, which was designed for 
the checking account-bookkeeping operations of banks, is a special- 
purpose machine. 

Decision making 

Another characteristic of electronic data processing equipment 
is its ability to make decisions. A typical decision is the comparison 
of two numbers to determine which is the greater. Another is the 
selection of one course of action from several alternatives. This 
decision-making ability results from the logical operations included 
in the instruction code. 

Electronic data processing equipment can make only predeter- 
mined decisions. This means that every decision that the equipment 
makes must be thought of previously by a human, and that the 
equipment has to be directed in the method of making the decision. 



Electronic data processing systems 101 

Electronic data processing equipment cannot make decisions for 
which it has not been instructed. For example, if the equipment is 
to choose among alternative courses of action on the basis of a prior 
computation, then the equipment has to be instructed not only on 
how to make the selection but also what to do after the selection has 
been made. 

ORGANIZING TO CENTRALIZE MACHINE DATA PROCESSING 

Among the most fundamental problems is the premise under 
which systems analysis is conducted. If a computer or punched card 
installation is used as the system processor, it is safe to say that data 
processing is top management business. It is fundamental to any 
successful centralization of equipment that management be well 
informed on the importance of data processing. Thus, top manage- 
ment education becomes a necessary building block in centralizing 
data processing. 

Management in one way or another sets the tone in which an 
organized attempt at systems analysis will be received. In addition, 
it is Management that will stipulate the relationship between the 
systems group and the departments being studied. Here are some 
of the possible ways in which the problem might be attacked: 

A. Without professional aid: 

(1) Departments will design their own systems. The rationale 
most frequently used is, "Who understands the problem 
better?" In this situation, problems are solved within a very 
limited frame of reference. Departments tend to see their 
own problems much magnified and out of perspective with 
the organization as a whole. The resulting systems (or more 
properly, methods) become data processing "islands" be- 
cause they are frequently only loosely connected to their 
adjacent areas. Integration and economy of operation can 
seldom be raised as critical organization problems because 
everyone is playing the part of systems designer in his own 
area of activity. 

(2) Management will allow a certain group to document sys- 
tems. In this instance, the accent is on developing proce- 
dures based upon existing practices. This is an improvement 
over the first alternative, since any review of a practice about 
to be adopted as department policy will have some benefit, 



102 Electronic data processing systems 

even when this review has no authority behind it. Proce- 
dures developed by departments for their own use may be 
too specialized to get a critical review; a Drawing Room 
Manual, for instance, would be a tremendous asset to any or- 
ganization having a multi-section engineering department. 
A quality control department might find a requirement for 
writing a Quality Control Manual to document the rules 
under which quality control engineers would reject mate- 
rials or workmanship in receiving, in process, in final as- 
sembly, in testing, or in shipping. In both of these examples, 
the review rendered by a documenting group may not be 
satisfactory. If the practices being reviewed are technical 
and mainly concerned with departmental methods, the bal- 
ance of the organization may be only superficially affected. 
As soon as these practices become reflected in interdepart- 
ment relationships, the need for surveillance will increase. 

B. With equipment and / or professional aid: 

(1) Management will provide equipment for data processing 
and some assistance in converting business systems to ma- 
chine systems, but will require using departments to provide 
their own operators and system designers. This is the 
method referred to as open shop; the equipment stays in one 
location, frequently without operators or technical experts, 
and is not aggressively used to provide service to staff. This 
method of operation has been successfully applied in engi- 
neering departments where a small scale computer is used 
to solve a narrow segment of mathematical problems and 
the conversion to machine language is easily learned. This is 
a costly method by comparison with the alternate possibili- 
ties of providing operators or technical assistance in prob- 
lem conversion, or both. Note that this situation is like A. 2. 
(page 101) in that outputs are mainly utilized in a limited 
number of organization units. 

(2) Management will create a data processing function report- 
ing to some existing organization unit which is a prime user 
of these services. This is frequently done, with varying suc- 
cess. In municipal government, the trend is to set up data 
processing under the treasurer or controller. In industrial 
organizations, the controller or chief finance officer fre- 
quently has prime responsibility. In both these instances, 
the priority of work on the data processor becomes a major 
issue. A second problem arises out of the anomaly that a 
controller may be asked to provide many non-accounting 
functions to users of equipment outside his department. 



Electronic data processing systems 103 

There is an organizational issue which invariably arises 
when a system analysis effort has been appended to the data 
processing group; some department managers will resist in- 
trusion by another department in resolving their problems, 
since it appears that someone at their own organization 
level (or below), is claiming jurisdiction and diluting the 
manager's autonomy. 

This problem can be solved by management's recognition that 
data processing is one of the largest, single, indirect costs of business. 
Management's education in data processing should stress the best 
organizational means of utilizing data processing. Here are some 
of the salient issues for top management: 

A. Leadership of centralized data processing means a closed shop 
where three fundamental services must be provided (see Figure 
6-1): 





DATA 

PROCESSING 
DEPARTMENT 
































SYSTEMS ANALYSIS 
SECTION 




APPLICATIONS 
SECTION 




OPERATIONS 
SECTION 


PRELIMINARY SURVEY 
DATA ACCUMULATION 
FLOW CHART 
FORM DESIGN 
PROCEDURES DESIGN 
FORMS CONTROL 
PROCEDURES MANUAL 
ORGANIZATION CHART 
ADMINISTRATIVE METHODS 
IMPROVEMENT STUDIES 
SYSTEM POSTULATION 


CONVERSION PLANS 

BLOCK DIAGRAMS 

PROGRAMMING 

CODING 

TECHNICAL ANALYSIS 


DATA PROCESSING SCHEDULE 
EQUIPMENT OPERATION 
SYSTEM OUTPUT 



BUSINESS SYSTEMS ANALYST 
OPERATIONS RESEARCH 
ANALYST 



COMPUTER ANALYST 

PROGRAMMER 

CODER 



COMPUTER OPERATOR 
TABULATING EQUIPMENT 

OPERATOR 
KEYPUNCH OPERATOR 
CONSOLE OPERATOR 



Fig. 6—1. Functional organization and job classifications associated 
with a typical electronic data processing department. 



(1) The design of data processing systems by business or engi- 
neering-trained systems analysts. 

(2) The conversion of data processing systems by specialists 
trained in the electro-mechanical or electronic features of 
equipment. 



104 Electronic data processing systems 

(3) The operation of equipment by personnel who are capable 
of obtaining maximum up-time (working time). 

B. Effective working relationships demand that Centralized Data 
Processing be managed by a man at the executive level: 

(1) This man must be no lower organizationally than the high- 
est using department. 

(2) The person to whom this manager would turn to obtain 
guidance or resolve a dispute must be at least one level 
higher than the disputants. 

(3) The delegation of authority to the manager of this group 
should come, where possible, from corporate management. 

C. The organization role of Centralized Data Processing must be 
clearly defined by top management; the objectives must be 
stated when authority and responsibility are established: 

(1) The organization must be made aware of Management's 
backing. 

(2) The organization should be told why it must accept the 
closed shop and its concomitant of systems analysts who will 
have a voice in the internal operations of individual depart- 
ments. 

(3) The central data processing group must feel it has both au- 
thority and responsibility to execute the tasks before it. 

D. Centralized Data Processing is best implemented by non-users 
who operate under strict budget provisions: 

(1) The non-user is objective and less suspect in matters such 
as job priority; the non-user will respond to a conflict in the 
data processing schedule differently than a user who will 
tend to put heavier emphasis on his own problems. 

(2) The budget for Centralized Data Processing should be ob- 
tained by assessing the needs of each division (department, 
section, group, and so on) and burdening them accordingly. 
Under these circumstances, where the users are paying for 
service, they will insist upon the central group performing 
their role, and cooperation will be a minor problem. It is 
more likely that there will be a long list of problems waiting 
for the systems analyst's attention. 

a) The data processing staff members should be delegated 
to their assignments according to a budget. Systems ana- 
lysts should work as close to the problem area as possible, 
and not "commute" to their assignments. A man assigned 
to "Department 10, Building 31," should have his desk 
and phone there; he must identify with the systems pur- 
chaser, without becoming involved in line responsibility 
of any kind. Contact with the central group should be 



Electronic data processing systems 105 

through frequent meetings to review work. This places a 
burden on the executive who leads the central data proc- 
essing group to provide esprit de corps and group 
identity. 

Centralization of data processing, itself, may require some justifica- 
tion. The economics of centralization are easy to demonstrate. It 
is desirable, therefore, to expose the criteria on which this judgment 
has been based to those who must accept the consequences of the 
decision. 



MAKING A FEASIBILITY STUDY 

The term feasibility has taken on the interpretation of possible 
or probable which, in the data processing context, is erroneous. Ob- 
viously, all of the computers analyzed in Chapter 7 could do a pay- 
roll or a mathematical problem. However, the resultant costs, time 
in process, and so on would vary widely. Likewise, the ability to 
integrate a given system into its related subsystems would vary 
widely, depending upon the choice of equipment. 

The term feasibility in the possible or probable context would be 
used mainly by manufacturers of data processing equipment. And 
in this claim they are no doubt right; each major manufacturer of 
equipment has a system capable of handling almost any known 
business problem. Even an exceptionally complex military problem 
such as the LOGBALNET of the Air Force, was finally mastered 
with a special system configuration which has over fifty tape drives 
of which twenty-three are on-line with a 700 series IBM computer. 

However, from the equipment user's standpoint, feasibility has 
come to mean something other than possible or probable. This en- 
lightened view places the emphasis on the following: 

(1) Analysis of equipment system alternatives in the light of data 
processing requirements. 

(2) Relationship of current data processing loads and equipment 
requirements to future requirements, based upon certain 
growth factors and anticipated improvements in equipment de- 
sign. 

(3) Consideration of system utilization versus cost to operate, using 
a number of rules. One such rule might be, "No system shall be 
considered a candidate for this company unless its utilization is 



106 Electronic data processing systems 

equal to, or less than, 40 per cent of first shift capacity for all 
applications which have been earmarked for conversion in the 
opening phase of implementation." 

(4) Analysis of programming costs and ease of conversion between 
alternate systems. 

(5) Analysis of facilities requirements of alternate systems. 

(6) Comparison of manufacturers' services (training, maintenance, 
programming, and so on) and "free" time, in a given installa- 
tion. 

Note that the accent here is on comparing, as nearly as possible on 
an equal plane, the equipment attributes. This, of course, places a 
burden on the user to describe his data processing requirements 
completely, so equipment manufacturers will be bidding on the 
same level. 

In addition, the feasibility study places the accent on systems 
design. Although there is no rule, a part of the potential savings 
sometimes claimed for mechanical or electronic equipment, is not 
entirely due to equipment. Systems redesign, alone, can sometimes 
pick up a part of the total savings to be enjoyed by converting to 
equipment. So there actually are arguments for continuing the pres- 
ent method (see Figure 3-1, p. 42). If a company can obtain a part of 
its total savings by system redesign alone, the cost of equipment begins 
to loom very large. The truth is that, frequently, equipment becomes 
the catalyst and either management and /or operating personnel 
will accept changes in the light of a new systems processor that they 
would reject if only the system were being revised. 

Equipment manufacturers have done a great deal to provide 
practical, economical, data processing systems in areas where iso- 
morphism does not take over. Payroll is a good example. Solutions 
to engineering problems such as earth removals, grade computa- 
tions, traverse closures, and so on have also reached a high level of 
efficiency because of the work of equipment manufacturers. How- 
ever, in areas like production control, material control, budget- 
ing, and so on, where the systems requirements are dictated in part 
by company size, the product, and the systems purchaser, equip- 
ment manufacturers can only suggest how the job has been done 
in certain specific instances. The burden is ultimately placed upon 
the user to determine if he can use a generalized system with certain 



Electronic data processing systems 107 

minor changes. The alternative is to undertake the study and de- 
sign of his own systems. 

A feasibility study to determine the usefulness of a computer in 
data processing has become a major undertaking. From the stand- 
point of elapsed time, study costs, and impact on the organization, 
company managements have thought twice before moving ahead. 
And yet, there is no suitable alternative in the majority of cases, 
something which will become more apparent after reading the out- 
line which follows on page 112. It will also become apparent that 
the properly oriented feasibility study is implemented by the sys- 
tems approach. 

Ostensibly, the feasibility study determines whether or not it is 
economically desirable to employ electronics in data processing. A 
study of this kind may require as little as six man-months, or take up 
to six man-years. Although the time dimension is very flexible, the 
content of such a study is generally the same. 2 

Broadly speaking, the feasibility study will undertake to examine 
the existing data processing operations and will determine which of 
these are susceptible to an electronic processor. Susceptibility will 
be based upon volume and frequency of processing, costs, and man- 
power required. Those tasks which are found susceptible in terms 
of some list of characteristics agreed upon by the study team will be 
costed to arrive at a total cost for the existing system operation. 
Despite the closely interwoven nature of business systems, the indi- 
vidual system costs must be isolated for future comparison with pro- 
posed electronic data processing costs. 

The study team will generally postulate a generalized system, on 
the assumption that a detail system design will not be appropriate 
until the data processor has been selected. Such a generalized sys- 
tem has limited validity, since it cannot take advantage of the special 
characteristics of any one equipment system. This is especially im- 
portant if evaluation of the proposed system results in marginal 
cost advantages. In this connection, intangible costs must sometimes 
be carefully evaluated. Although there may be no measurable dollar 
savings for faster or more accurate reporting, the addition of many 
intangible benefits sometimes serves to balance the marginal savings 
in the direction of an order for equipment. 

2 Sec Marxson and Company, Chapter 12. 



108 Electronic data processing systems 

Using the postulated electronic system as a guide, there are two 
courses open to a company: 

(1) Prepare general specifications (see Figure 6-2 for an example of 
a basic data collection form) of the data processing system 



DATA SPECIFICATION SHEET REFERENCE 

Title 

Data Processing 

Purpose 

Physical Form Method of Preparation 

Sources 

Dispositions 

Quan used Avg. Max Over Period of 

Quan of Records/ File Avg Max. Freq. of Reference 


Item Group 


Item Group Summary 


Char/Item Group 


Tot. Separate Groups 


Average 


Maximum 


par Record 


per File 


Alpha 


Num. 


Alpha 


Num 


Aver 


Max 


Aver. 


Max 


















































































































































Internal Sequence 


Item 
Group 


Item 
No. 


Item Description 


Quantity Chars. /Item 




Average 


Maximum 




Alpha 


Num 


Alpha 


Num. 




















































































































































































Remarks 



Fig. 6-2. 



Electronic data processing systems 109 

which has been studied. Assemble this in books together with 
sample forms, reports, flow charts and so forth, plus a set of rules 
or criteria explaining your requirements and the basis upon 
which the system will be selected. Turn over this data to at least 
three manufacturers of equipment asking for a proposal within 
thirty to sixty days. Equipment manufacturers are staffed to do 
this at no charge. After proposals are returned, the user of 
equipment must undertake equipment evaluation, the logical 
comparison of advantages and disadvantages of one system over 
another. 

(2) Utilizing the specifications which have been prepared, assemble 
the manufacturers' literature, making an equipment evaluation 
without the benefit of manufacturers' proposals. 

In either case, note that a technically oriented group must step in 
to do equipment evaluation (see Chapter 7). This task is becoming 
more and more difficult as time goes on, because of the variety and 
complexity of systems which are available. Each company must de- 
termine how to supply the requisite skills in this specialized field. 
It is appropriate to divide the feasibility study into two parts: (1) 
the preliminary study to determine the coarse economics of a com- 
puter versus the existing method of data processing for a given com- 
pany, and (2) the generalized system study (feasibility study) result- 
ing in a system postulation which is assembled as an invitation to 
manufacturers of equipment to bid. 



QUESTIONS TO BE ANSWERED BY A PRELIMINARY SURVEY 

Prior to making heavy investments of personnel and time in ex- 
tensive, detail analyses, a preliminary survey will show the eco- 
nomics of converting to a computer. If a preliminary survey reveals 
the high probability of a dollar payoff, then further study would be 
warranted. Following are some typical questions to be answered in 
a survey of this type: 

(1) Does the company have computer applications? 

(2) If so, what are they? 

(3) Will there be a dollar payoff associated with these applications? 
What is the nature of the payoff? Will there be reductions in 
out of pocket expense? 



110 Electronic data processing systems 

(4) Will the dollar payoff be large enough to make it practical to 
sustain the costs of converting to electronics and of maintaining 
a computer center? 

(5) What should the priority of study be in converting applications 
to a computer? 

(6) What will be the duration of time and how much will it cost to 
prepare an invitation to bid? 

(7) Why use electronics in preference to some other method of data 
processing? 

Here are some typical questions one might ask of either the exist- 
ing or proposed (postulated) system: 

(1) Does the proposed system provide clear-cut improvements over 
the existing system? What are they? Are they measurable? 

(2) Can the existing system be improved substantially by more in- 
tensive study? Will such improvements invalidate the bulk of 
the proposed savings of an electronic system? 

(3) Does the proposed system call for generally acceptable organi- 
zational changes or changes in method? If some changes are un- 
acceptable, will modification restrict the dollar benefits to be 
gained? 

(4) Will the proposed system savings be effected by reductions in 
manpower? Will there be specific curtailment of expense in the 
proposed system? Will open requisitions for manpower be can- 
celled? Will personnel be transferred? Will the employee quit 
rate be sufficient to provide savings predicted by the proposed 
system? (See Chapter 8.) 

(5) Does the proposed system perform more quickly, more effi- 
ciently, or more accurately than before? Is this measurable? 

(6) Does the proposed system provide outputs which are perma- 
nently required? Are they currently available? Are they un- 
available because of manpower, time, or communications limi- 
tations? Are they unavailable because they are beyond manual 
computational abilities? 

(7) Is the proposed system practical and operable within the per- 
sonnel limitations of the organization? Is it similar enough to 
other systems so its validity in operation is unquestioned? 

(8) Are the training, indoctrination and orientation requirements 
of organization personnel within some reasonable scope? Are 
appropriate personnel available within the organization? Is the 






Electronic data processing systems 111 

labor pool for personnel sufficiently deep to insure system con- 
tinuity? 

(9) Do system requirements impose excessive or severe demands on 
personnel? Can the system operation be simulated in advance to 
determine its reliability, and its effect on the organization? 

Here is a final checklist that might be directed at the generalized 
system study prior to asking equipment manufacturers to prepare 
proposals: 

(1) Is the system requirement satisfied by the postulated system? 

(2) Are the data handling steps described adequately as they are 
interrelated? Procedurally? In time? 

(3) Are data inputs described as to source, format, method of 
transmission, and volume per unit of time? 

(4) Have the data storage and retrieval requirements been de- 
scribed as to the kinds of data? The nature and size of each 
item? Items per record and records per file? The volume of 
data stored or filed? The access requirements? Interrogation 
and answer requirements? Methods of file storage and main- 
tenance? 

(5) Have the data processing requirements been described in 
terms of the kind of processing that will be required? The 
speed with which outputs must be made available? The vol- 
umes to be processed? On-line requirements? 

(6) Are the outputs described as to format and content? Are the 
quantitative characteristics clear (alpha-numeric content, vol- 
ume, number of copies, and so on)? Have the time require- 
ments on output availability been described? Have peak load 
requirements been expressed? 

(7) Has the use of outputs been clearly stated? Are they for the 
preparing agency control only? Are they forwarded under a 
priority plan? Is there some urgency factor? Are the outputs 
incorporated in other outputs or are they final in nature? 

(8) Have the integration requirements been stated? Have the me- 
chanics of flow charting clearly illustrated how integration is 
to be achieved? Have system inputs been correlated with multi- 
output requirements? 

(9) Has the human role been defined? Do humans continue to file, 
sort, route, find, calculate, and copy. At what points do human 
judgment and interpretation enter the system? What are the 
control mechanisms over humans? 



112 Electronic data processing systems 

(10) Are the system attributes concerning feedback and control 
stated? Are systems fully stated as modules with all of the spe- 
cific attributes? 

(11) Have any special purpose requirements been stated? If so, 
what is the impact of the special requirement on the system 
as a whole from the standpoint of cost, delivery, maintenance, 
and so on? 

The actual feasibility study may take many forms based upon 
the fundamental premises which have been stated. The outline 
which follows, is a fairly well established and successful pattern 
which can be utilized in most cases. 



OUTLINE OF HOW TO CONDUCT A FEASIBILITY STUDY 



I. Business analysis and problem statement 

A. Organization Charts 

1. By area 

2. By activity 

3. By operation 

B. Statements of Objectives 

1 . By area 

2. By activity 

C. Analysis of Present Operating Procedures 

1. Data gathering 

a. Interviewing and note-taking 

b. Organization of data 

2. Data presentation 

a. Sample documentary forms 

b. Lists of files and records required 

(1) Functions 

(2) Size 

(3) Access requirements 

c. Communications requirements 

(1) Within area 

(2) Between areas 

d. Preparation of flow charts 

(1) Index 

(2) Area by activity — quantitative data 

(3) Activity by operation — quantitative data. 



. 






Electronic data processing systems 113 

D. Cost Estimates 

1. Personnel 

2. Operating costs (direct) 

3. Overhead charges (indirect) 

4. Equipment 

5. Facilities 

II. Data processing system design 

A. Analysis of Present Data Processing Procedures 

1. Area by activity communications 

a. Inputs 

b. Outputs 

c. Processors 

d. Feedbacks 

e. Controls 

2. Activity by operation communications 

a. Inputs 

b. Outputs 

c. Processors 

d. Feedbacks 

e. Controls 

B. Description of Postulated Electronic Data Processing System 

1. Operational flow charts 

2. Rank equipment 

a. Inputs 

b. Operations (processes) 

(1) Sequential 

(2) Decision 

c. Outputs 

3. Changes from present business systems flow charts — Con- 
solidation and elimination of files and operations 

4. Runs to be performed 

a. List and description 

b. Block diagram 

(1) Inputs 

(2) Processors 

(3) Outputs 

(4) Feedbacks 

(5) Controls 

5. Files and records to be maintained 
a. Description 



L 



114 Electronic data processing systems 

b. Quantitative data 

c. Access requirements 

6. Recommended forms of inputs 

7. Reports to be prepared 

8. Communications considerations 

a. Integrated data processing 

(1) Within area 

(2) Between areas 

b. Facilities required 

9. Time estimates and schedule (coarse figures) 

a. Machine running time 

b. Peripheral equipment 

c. Over-all time schedules 

10. Cost estimates (coarse figures) — prorated by activity 

11. Operational features of significance in the specific applica- 
tion. 

12. Over-all appraisal. 

C. Determination of Specific Equipment to be Considered 

1. Types of equipment and alternatives 

2. Manufacturers and models to be investigated 

D. Interim Feasibility Study Report 

1. Brief description of activity 

2. Description of the data processing system (summary of Figure 
6-2) 

3. Conclusions: Over-all appraisal of the suggested system 
a. Advantages and disadvantages 

(1) Economic 

(2) Operational 

III. Equipment evaluation to implement a postulated data 
processing system 

A. Briefing on Postulated System for Specialists in Specific Equip- 
ment by the Data Processing Engineer Assigned to Each Area 

B. Equipment Comparison Data Presentation by Each Specialist 

1. Flowcharts 

2. Specific equipment required 

3. Runs to be performed: 

a. List of runs (passes) 

b. Block diagram of each 

4. Time 

5. Cost 



Electronic data processing systems 11 s ) 



a. Equipment 

b. Personnel 



C. Consolidated Report by Data Processing Engineer Assigned to 
the Area — Specific System and Equipment for Each Manufac- 
turer Involved 

A GLOSSARY OF USEFUL TERMS USED IN 
ELECTRONIC DATA PROCESSING 

access time: The time required to locate a word in storage and transfer 
it to the processor. 

accumulator: A register in the processor that holds sums, products, 
and so on; also used for comparing data. 

adder: The component of a machine that does arithmetic; it may or 
may not serve simultaneously as an accumulator, and it may or may 
not add by "counting." 

address: The designation of a storage location. 

address part of an instruction: The portion of an instruction that 
tells the machine where to find the data that is to be used in an opera- 
tion, or where to store a result. 

address system: A term that designates the number of addresses con- 
tained in a single instruction. 

binary number system: A number system with only two digits, and 1, 
in contrast to the decimal system which has 10 digits. All current 
machines are either decimal or binary. Arithmetic is done in a differ- 
ent manner with the binary system than with decimal. 

bit: A bit is the least amount of information that can be represented as 
zero or a one, as represented by a hole in a card — or no hole; or as 
represented by a plus or minus electrical charge. Bits are grouped in 
pattern to represent coded characters. 

branching: Automatic selection of appropriate alternate instructions. 

buffer: A register that can operate at (usually two) different speeds, 
thus permitting processing of data at high speed while simultaneously 
reading input and /or writing output. A temporary storage unit. 

control: The section of a data processing machine that controls its 
operation. Most current machines are either manually controlled 
through a keyboard or have a wired control panel or operate by a 
stored program. 



116 Electronic data processing systems 

feedback: The technique of returning selected portions of the results 
produced by a processor to its own control section for the purpose of 
self-supervision or modification of further processing. 

input: Information delivered to a machine; also, the section of the 
machine that "reads" this information. 

instruction: A word which is sent to a register in the control section of 
a machine, where it causes the machine to perform a particular opera- 
tion. An instruction will consist of an operation code and one or more 
addresses. If all instructions for a particular machine have just one 
address part, then the instructions and the machine are referred to 
as single address. Similarly, a three address machine will have three 
addresses and one operation code to each instruction (see Address 
System). 

logical components: The sections of a processor capable of making 
comparisons, testing for algebraic signs, identifying zero or non-zero 
results, and so on. 

memory: The section of a machine used to store data and instructions. 

microsecond: One millionth of a second: .00001 seconds. 

millisecond: One thousandth of a second: .001 seconds. 

modification: The ability to alter the normal sequence of instructions 
by branching or changing the command or address portion of instruc- 
tions. 

operation code: The portion of an instruction that tells the machine 
what operation to perform. 

output: Information delivered by the machine; also, the section of the 
machine that "writes" this information. 

processor: The section of a machine that performs arithmetic, makes 
logical decisions governing control, and processes the data in storage. 

program: A list of instructions to the processor. 

read: A command instructing the machine to accept data from a specific 
input unit and place it in memory. 

record: A group of related words. 

record storage unit: An intermediate memory unit used to facilitate 
transmission of records between dissimilar units. 

register: Any storage unit used for a specialized purpose. A register is 
usually designed to contain a fixed number of characters. 

storage: The section of a machine that stores data or instructions: the 
memory section. 



Electronic data processing systems 111 

storage location: The address of any specific area of memory. In a fixed 
word length machine it refers to the address of a specific word. In a 
variable word length machine it refers to the address of a specific 
character. 

stored program machine: One which is controlled by having an appro- 
priate list of instructions stored in the machine just as data is stored. 
The machine refers to these instructions in a specified sequence, ex- 
ecuting each before examining the next. 

word: A unit of information; a piece of data equivalent to a field in 
punched card terminology. 

word length: The number of characters in a word. 

fixed word length: Storage designed to contain a specific number of 
characters for each addressable location. 

variable word length: Storage designed to store words of any size 
in consecutive locations to make more efficient use of memory. 

write: A command to a machine to take information from memory and 
record it on some output media. 



7 



Evaluation of equipment 
systems 



INTRODUCTION 

New equipment and modifications of existing equipment are con- 
stantly enlarging the computer market. This section is not intended 
to embrace all equipment, but to provide a cross section of equip- 
ment available at this time. Direct contact with the manufacturers 
of electronic computers will enable the systems analyst to obtain 
complete information on available equipment. Manufacturers have 
a wide range of technical and non-technical data available, a large 
part of which is distributed free of cost to potential customers. 

Data on each computer are arranged under the following head- 
ings for easy comparison: 

General characteristics and information 

Number system 

Instruction system 

Storage (drum, core, magnetic tape) 

Input methods 

Output methods 

These equipment characteristics — and the more detailed tech- 
nical facts of these systems — would normally be explored in the 
evaluation of computers for use in a given installation. Equipment 
evaluation is the means by which these equipment characteristics 
are tested through the performance of specific data processing prob- 
lems. This requires not only the general type of knowledge which 
is contained in this chapter, but a detailed knowledge of equipment 
characteristics, construction (circuitry), codes, program routines, 
and so on. Equipment evaluation is the final test of the desirability 

118 



Evaluation of equipment systems 119 

of one computer over another, and it can only be undertaken by 
technically competent personnel. Competent selection of a com- 
puter will reveal the cross section of equipment most likely to pro- 
duce the proper input, output, and processing capabilities which 
the company's data processing system requires. Differences in ma- 
chine speeds and modes of operation will make the costs of data 
processing vary, since some computers will operate more efficiently, 
on a given problem than others. Not the least of the problems to be 
surveyed in equipment evaluation will be the ease of programming, 
a subject which will not be explored in this text. The costs and 
problems imposed by programming make this a major and some- 
times determining factor in the choice of a computer system. 

Four computers have been described in parallel in this chapter. 
They have been selected because they represent something of the 
current variety available in electronic equipment; there are many 
other computers available, and each has a number of special charac- 
teristics to recommend it. Even these examples have been super- 
ficially appraised because a full, technical description would be out- 
side the mission of this text. 

The objective in presenting this material is not to train techni- 
cians but to bring home the complexity of equipment evaluation 
and to stress that top management must take a technical point of 
view when it is dealing with electronic data processing. Machine 
data processing must be considered a level above manual systems, 
and, today, electronic data processing is at the highest level in this 
category. The use of computers imposes organization and system 
constraints not encountered in manual systems, something which 
will be readily seen by viewing the attributes of the four equipment 
systems that follow. 

DATATRON 220 1 

General characteristics and information 

This is a general purpose machine capable of handling a wide 
variety of business problems as well as scientific and engineering 

l Manufactured by the Burroughs Corporation, Electrodata Division. 



120 Evaluation of equipment systems 

applications. It is normally considered a large scale computer be- 
cause of its large memory capacity, high speed, and general price 
range. 2 It is a stored program machine, as are all of the other medium 
and large scale computers. This means that the individual steps of 
a complex process can be held internally, which allows the equip- 
ment to go from one step to the next without human interference. 
Because no two systems are necessarily made up of identical equip- 
ment, lease prices or purchase prices must be computed based upon 
the selected equipment system. Thus, the prices shown in this chap- 
ter should be accepted in a general way since they indicate only one 
system configuration. The "220" system leases for approximately 
$25,000 per month; its purchase price is approximately $1,250,000. 

The "220" system requires in excess of 60 kilowatts for its opera- 
tion. Temperature and humidity control are necessary. Approxi- 
mate system weight is 140,000 lbs. Space requirements for a full 
system would approach 1,200 square feet; additional space would be 
required for personnel, files, storage of cards, and so on. 

Analyst, programmer, operator, and maintenance training is pro- 
vided by the manufacturer (usually free). Some problem analysis 
is provided, although emphasis is on initial programming assist- 
ance. Extensive library routines are made available to customers, 
generally at no charge except for materials. Data processing facilities 
are available on an hourly or fixed fee basis for specific problems; 
these facilities are also available prior to the installation of a cus- 
tomer's machine for the checking of programs and running of 
sample problems. 

Number system 

The machine's number system is important since it indicates the 
way in which its instructions will be constructed, and the way in 
which data will be held internally. In the "220," as in other ma- 
chines, the number base is what is known as binary coded decimal. 
The number system is capable of showing decimal digits through 
9, plus 26 alphabetic characters and all other typewriter and special 
symbols. Alphabetic (alpha) data is carried internally by pairing- 
decimal digits. 

2 Price quotations on equipment of all manufacturers are as of April, 1959. 



Evaluation of equipment systems 121 

Instruction system 

The computer is operated by means of instructions, which, step- 
by-step, direct the operations which are to be performed. Instruc- 
tions are carried by computer words of a special type within the 
machine. A picture of a computer word for the "220" would appear 
as in Figure 7-1. One of these words is required for each instruc- 



Picture of Stored Numerical Information: 

(digit positions) 1234567890 



SIGN 



Decimal Digits 



Picture of Stored Alphanumeric Information: 

+ 1234567890 
( digit positions ) 



2 


4 


1 


8 


2 


8 


3 


8 


4 


4 


2 



Fig. 7-1. 

tion, and they are so constituted that they work sequentially. There 
are approximately 92 different commands (instructions) for the 
"220" system (Figure 7-2 is a picture of one of these instructions). 



+ 


(digit positions) 
1 2 3 4 5 6 


7 8 9 


SIGN 


CONTROL DIGITS 
1 1 1 


OPERATION 
CODE 

1 


ADDRESS 

1 1 1 



Fig. 7-2. 

In this command structure, there are special instructions which 
allow operations to be performed on partial word fields. Other 
special instructions permit the transfer of 1 to 100 words of informa- 
tion (called a block of information) from one part of internal stor- 
age to another. Another special instruction tests for relative magni- 
tude and transfers control of subsequent operation steps to one of 
four statuses: high, low, equal, or transfer. Through the special in- 



122 Evaluation of equipment systems 

struction called magnetic tape scan, the computer is able to locate 
blocks of data on magnetic tape which belong to a certain class. 

Storage 

This is a magnetic core machine. The normal system will contain 
from 2,000 to 10,000 words of storage (prices quoted are based on 
5,000 words), in increments of 1,000 words. Memory size is of basic 
importance, since it will act as a restriction on either the size of files 
to be stored or the number of instructions that can be stored, al- 
though these may not be accommodated in the same part of the sys- 
tem. The amount of storage required will be a direct reflection of 
the types and sizes of files to be stored. Core or drum storage is re- 
ferred to as internal storage since it is physically contained within 
the central processing unit of the system. The "220" system is one 
of the random access machines, since the instruction code permits 
direct (non-sequential) retrieval of any data stored in the memory — 
in any sequence. The average time to find any single piece of data in 
core storage is 15 microseconds (millionths of a second). 

Auxiliary storage is provided in magnetic tape units which are 
external to the central processor. Ten tape carriers can be placed on- 
line with the "220" system. Each tape reel is 3,500 feet long and con- 
tains data at the density of 416 decimal digits per inch. Thus, any 
single tape reel can carry 15,000,000 digits of data. Movement of 
data from the carriers to core storage or other tape carriers is ac- 
complished by means of a magnetic tape control unit, one of which 
is required per "220" system. The read- write heads of the magnetic 
tape storage unit will pass over 25,000 digits per second which is the 
equivalent of reading an entire 3,500 foot tape in less than six 
minutes. Recorded information can be found by searching in either 
forward or backward motion, each tape unit operating independ- 
ently of others. In order to update specific pieces of information, 
writing heads can selectively write over recorded data. 

All reading operations are done twice and compared. Imperfect 
tape areas are machine detected and automatically rejected for data 
recording. Many special error detection devices exist which are 
common to most computers. 

The following are the average times, including time to obtain 



Evaluation of equipment systems 123 
the instruction and operand from storage and time to do the opera- 



tion: 



Operation Time (microseconds)' 



Addition or subtraction 200 

Multiplication 2,095 

Division 3,985 

Transfer 125 



Input methods 

The ''220" employs the Cardatron system by which the relatively 
slow card handling machines are coupled to the high speed central 
processor via buffers (drums) or intermediate storage devices. There 
can be up to seven Cardatron Input devices per system (and seven 
Cardatron Output devices). One Cardatron Control Unit per "220" 
system is required. The Cardatron Input Unit transfers data to the 
Cardatron Control Unit from a card reader at the maximum rate of 
480 cards per minute, without interfering with the work of the 
central processor. The control unit is capable of operating at the 
rate of 15,000 digits per second. Transfer time from the Cardatron 
Input Unit to the central processor is 7.2 milliseconds (thousandths 
of a second). 

A seven channel paper tape is also used as input to the "220." Data 
is read by a Photoreader which will carry a 5 1/ 2 inch (40,000 charac- 
ters) or 7 inch (80,000 characters) diameter reel at a density of 10 
characters per inch. Tape can be read at the rate of 1,000 characters 
per second (100 inches per second). The Photoreader can do auto- 
matic translation, as well as make parity checks and terminate read- 
ing by word count or control data on the tape. 

The "220" system also accepts IBM punched cards which means 
that the total cost of a computer system utilizing this type of input 
must include key punches, key verifiers, reproducing punches, 
summary punches, and so on. Here again, data processing systems 
will dictate the combination of equipment necessary to meet the 

3 Millionths of a second. Other manufacturers express their average times in milli- 
seconds, thousandths of a second. Times have been shown as the manufacturer cus- 
tomarily expresses them. 



124 Evaluation of equipment systems 

system requirements. It is not uncommon to combine the equip- 
ment of two or more manufacturers in arriving at the equipment 
complement. 

The central computer contains a control console, one feature of 
which is a Supervisory Printer. This unit is a modified typewriter, 
and has a keyboard which enables the console operator to make 
various tests and special interrogations of the system. Likewise, it 
can receive data from the system, printing answers on a sheet of 
paper. This feature, common to most large scale systems, is valuable 
in error-detection, debugging and testing. This input is, of course, 
manual by means of a keyboard. 

Output methods 

The Cardatron Output Unit receives cards via the Cardatron 
Control Unit which controls card punching at the rate of 100 cards 
per minute, maximum. The Control Unit likewise monitors line 
printing at the rate of 150 lines per minute by means of a buffer 
drum which frees the central processor for its own work. Time to 
move from the central processor to Cardatron Output Unit is 10.0 
milliseconds. Card data is independently edited under electronic 
format control on both input and output. Control is via a stored 
program. 

The "220" system also utilizes a Paper Tape Punch which oper- 
ates off an 8 inch diameter reel. This unit operates at a speed of 60 
characters per second (6 inches per second feed), substantially slower 
than input rates previously quoted. To partly overcome this dis- 
crepancy, the "220" system can be set up to operate with Paper Tape 
Punches, Photoreaders and Character-at-a-Time Printers up to a 
maximum of 10 units per system. The Paper Tape Punch has an 
automatic alpha, bi-decimal translation capability (see Instruction 
System) as well as automatic parity generation and zero suppression. 

The "220" system contains one Character-at-a-Time Printer (also 
called the Supervisory Printer). This is an off-line device. Like a 
typewriter, it will print alphameric data at a rate of ten characters 
per second. It has carriage control, tabular stops, automatic zero sup- 
pression, and can be equipped with an off-line mechanical reader. 
In this arrangement, it can type data direct from punched paper 



Evaluation of equipment systems 12 s * 

tape at the rate of ten characters per second with carriage control, 
tabular stops, and so on controlled from paper tape rather than the 
central processor. 

The "220" system is capable of producing IBM punched cards as 
an output via reproducing punches or summary punches, at the 
speed of 100 cards per minute. Control is under the central processor 
and generally only one card output device may be operating at a 
time. Data pass through a card converter which acts as a buffer trans- 
lating machine code into card code, and up to eight, ten-digit words 
can be punched in a card. 

The "220" system, in addition, utilizes IBM Tabulators (402, 
403, 407) as an output device. Format control is via the printer's 
control panel; sequence control is from the stored program. Data 
pass through a buffer as with card output, and, again only one 
printer may be operating at a time on line. Line speeds are normal 
IBM printer speeds. 

Magnetic tape is an input-output device as well as being a storage 
device. The storage feature (external memory) is achieved when a 
basic file is put on line for updating. This, of course, is input to a 
data processing system while at the same time it is a holding device 
for information. After the file has been made current by deleting 
old data and adding new input data, it becomes output, and, as 
such, still retains its external storage character. 

HONEYWELL 800 4 

General characteristics and information 

This is a general purpose computer with a range of capabilities 
much like the Datatron 220. One significant design feature of the 
"800" is its all transistorized construction, which eliminates the 
bulk of the vacuum tubes normally used in computers. Not only is 
the system reliability thus improved, but, in addition, less heat is 
generated, less power is consumed, and less air conditioning is re- 
quired. This system is in the intermediate class — between medium 
and large scale systems. Its approximate lease price would be $20,000 

4 Manufactured by DATAmatic, a division of Minneapolis-Honeywell Regulator 
Company. 



126 Evaluation of equipment systems 

per month and its purchase price would be about $1,000,000, de- 
pending upon system configuration. The "800" can assume the 
price and memory characteristics of a medium, intermediate, or 
large scale system because of certain flexible design features. This 
has been done by providing input-output equipment which is 
designed on a standard and high speed basis giving two distinct 
operational modes. The high speed devices are more costly, and, 
hence, the total costs of a system with high speed devices would be 
substantially larger than the standard equipment. It would be logi- 
cal to turn to high speed equipment if, for instance, the output re- 
quirements were such that standard printing rates (150 lines per 
minute) would be too slow for the requirements of the data process- 
ing system. This system shows a design relationship in many ways to 
the large scale DATAmatic 1000 of which there are only six in the 
world. The dependency which some computer manufacturers place 
upon IBM to supply all input-output peripheral equipment is not 
true in the case of this manufacturer. 

The ''800" system requires 25 KVA for its operation. Air cooling 
is built into the system and reduces the need for special facilities 
treatment. Room size for an intermediate scale system would be 
about 1,000 square feet, not including space for personnel or records 
storage. The system weight would be about 12,000 pounds, depend- 
ing upon exact numbers and pieces of equipment. 

The manufacturer offers a variety of training courses ranging 
from executive seminars of one day through operator's courses of six 
weeks, all free of cost. Complete maintenance by resident DATA- 
matic personnel is included in all lease contracts. Maintenance of 
purchased equipment may be arranged by a separate contract or by 
training of customer personnel at no charge. The manufacturer is 
establishing computing service centers at strategic points in the 
United States where sample problems and customer debugging can 
be undertaken. 

Number system 

This system operates on binary or binary coded decimal system 
internally. Externally, the "800" will operate in octal, hexadecimal 
or decimal. Internally, the number system is capable of through 9: 



Evaluation of equipment systems 127 

externally, it can show 10 decimal digits or 16 hexadecimal digits. In 
addition, it can handle 26 alpha characters and 20 special symbols 
which are stored in a six bit binary form. 

Instruction system 

The "800" word is 54 binary digits, including 6 check digits; 
this is equivalent to an 11 decimal plus sign or 12 decimal digit 
word. Figure 7-3 is an example of an "800" numeric Avord; Figure 



1101 


0101 


0011 


0001 


0010 


0101 


0000 


1001 


0110 


1000 


0110 


1000 


+ 


5 


3 


1 


2 


5 





9 


6 


8 


6 


8 



Fig. 7-3. 

7-4 is an example of an "800" alphabetic word; Figure 7-5 is an 
example of an "800" mixed (alphameric) word. Note that all words 



110110 


011011 


110010 


011011 


010010 


110100 


010011 


100010 


w 




s 




B 


u 


c 


K 



Fig. 7-4. 



010010 


110100 


010011 


100010 


0010 


0101 


0000 


1001 


0110 


1000 


B 


u 


c 


K 


2 


5 





9 


6 


8 



Fig. 7-5. 



contain a maximum of 48 binary digits, in combinations of four 
and six bit blocks (check digits not shown). 

This is a three-address machine (as contrasted with both the 
Burroughs and IBM equipment). An instruction may lead to the 
next instruction in normal sequence, to a temporary departure 
from and automatic return to normal sequence, or to a complete 
change to a new sequence. In addition to three addresses, every 
instruction specifies either of two sequencing counters as the source 
of the next instruction. Both automatic and programmed sequenc- 



128 Evaluation of equipment systems 

ing is possible. There are 51 basic instructions; they are structured 
as in Figure 7-6. The 54 bit word has great system design flexibility. 



OPERATION 
CODE 


ADDRESS 1 


ADDRESS 2 


ADDRESS 3 


CHECK 
DIGITS 


1 >12 


1 >12 


1 +12 


1- — >12 


1 >6 



Fig. 7-6. 



The operation code specifies the instruction to be performed and 
(depending on the instruction category) such information as the 
source of the next instruction, fixed or floating point arithmetic, 
peripheral device involved, partial mask address, and memory loca- 
tion or special register option for each of the three address groups. 
The structure of the address groups depends on whether a memory- 
location or a special register is addressed and whether addressing is 
direct, indirect, or indexed. The following paragraphs illustrate 
some special instructions that facilitate operation. 

Multiple transfer is an instruction designed for indirect address- 
ing. It allows the programmer to assemble or to distribute informa- 
tion within memory and is particularly useful in sorting or matrix 
computations. 

Compute orthocount is an instruction which causes the auto- 
matic generation of orthotronic count information accompanying 
every record written upon or read from magnetic tape. This in- 
formation is used to check both writing and reading and provides 
a means for reconstructing damaged data — in other words, it pro- 
vides a virtually complete internal system to protect the files. 

Shift and select is an instruction designed for use in data process- 
ing and logical computation areas. It automatically selects a course 
of action based on information within the item being processed, for 
example, in the handling of transaction codes. 

Distributed read-write causes the computer to read a record into 
memory and distribute the items comprising the record to pre- 
assigned locations, or to assemble a record from such distributed 
items for writing. This ability is included in the peripheral read 
and write instructions. 

Simulator instructions are a group of instructions, virtually 
unlimited in number, which can be defined in advance by the pro- 



Evaluation of equipment systems 129 

grammer, who can subsequently call for a desired subroutine, com- 
prising a number of instructions, with a single instruction. 

Whenever information is transferred within the Honeywell 
"800," every word is checked for accuracy. This includes transfers 
between peripheral devices and memory, memory and tape, and all 
transfers within the central processor. Similar checking verifies all 
arithmetic and control operations. A built-in marginal checking 
system permits convenient, periodic checks of component perform- 
ance levels. Orthotronic control is a unique automatic data-pro- 
tection technique. At the end of each record, Orthotronic check 
numbers are added to each information channel. Words which 
might become lost or garbled on magnetic tape are internally re- 
generated by Orthotronic procedures, which eliminates costly and 
time-consuming manual corrections. 

Storage 

The "800" is a stored program machine. Unlike some computers, 
any location can store any valid "800" word. Access to or from any 
memory location is in parallel. Storage is of the magnetic core 
variety in increments of 4,096 words (8,192, 12,288 or 16,284 are 
also available). The average access time in the standard configura- 
tion is six microseconds. 

External storage is possible in from 1 to 64 magnetic tape units 5 
per "800" system. Tape speed is 120 inches per second, but rewind 
can be done three times as fast. Tape control is via the stored pro- 
gram for on-line operations and manually from the console for off- 
line operations. Tape reels are 2,500 feet and store 1,666,666 "800" 
words or 20,000,000 decimal digits. Information is transferred at the 
rate of 96,000 decimal digits per second per magnetic tape unit. Up 
to eight units may be reading and up to eight other units may be 
writing at any time. Each group of words written or read as the 
result of a single instruction is called a record. Records may vary 
from 2 to 400 words in length and are separated by gaps of 2/ s inch. 

5 Not to be confused with the DATAmatic "1000" tape units, the "800" tape units 
are y 4 inch mylar (plastic) like the tape units of other manufacturers. The "1000" 
tape units are 3 inch mylar and require their own special tape carriers which are not 
compatible in the "800" system. 



130 Evaluation of equipment systems 

Writing is done with the tape moving in the forward direction, but 
reading may be done with the tape moving in either direction. 

A magnetic tape switching unit is available for selecting on-line 
or off-line connections. Peripheral control devices provide the nec- 
essary buffering and conversion for all input-output units. Up to 56 
inquiry stations can be included for remote random interrogation 
of magnetic tape files. Thirty thousand three address instruction- 
operations per second are possible, which makes the "800" among 
the fastest computers. The following are average times for basic 
logical and arithmetic operations: 



Operation 


Time (milliseconds) 


Addition or subtraction 


.024 


Multiplication 


.150 


Division 


.312 


Transfer 


.018 


Branch 


.024 


Shift 


.036 (avg.) 



The Multi-Program Control Section permits up to eight pro- 
grams to function simultaneously. Each program proceeds inde- 
pendently of the others without the need for special programming. 
The Traffic Control Section monitors up to 16 input-output trunks 
and effects the necessary connections at the proper times between 
these trunks and the central processor. Virtually any desired com- 
bination of input devices, output devices, and magnetic tape units 
can be connected to these parallel trunks via appropriate control 
units. 

Input methods 

The "800" provides five flexible means of input. The first of these 
is the Flexowriter which is the console typewriter. Not only is a 
written record of the interrogation made at typewriter speeds, but 
a punched paper tape by-product is simultaneously prepared. This 
input device is used in debugging, error detection, and sometimes is 
used to insert starting or other special data directly to memory. 

Punched paper tape itself can be the input device operating at 
typewriter speeds (ten characters per second). It fills much the same 



Evaluation of equipment systems 131 

role as the typewriter and, of course, is too slow to be involved in 
system processing. There are two punched paper tape readers avail- 
able that can be used to introduce data into the system. These oper- 
ate at 200 and 1,000 characters per second, respectively, and are 
known as high speed readers. 

There are, in addition, three Magnetic Character Readers, which 
are capable of reading documents imprinted with magnetic ink. 
These operate at 750, 900 or 1,500 documents per minute and add a 
new dimension to information processing. 

The final input method is via IBM punched cards. Here again, 
great versatility is provided in card readers that operate at 240, 650 
or 900 cards per minute, either on-line or off-line. Information 
from each card is read twice and the readings compared. Each card 
is converted to ten words in memory. Cards containing invalid 
punches may be rejected or converted with an invalid-punch signal. 
Card columns may be rearranged or deleted by means of a control 
panel. Orthotronic control information is automatically generated 
and attached to each record by a card reader. Note the effect on over- 
all system cost of high speed devices. 

Output methods 

The "800" system provides five methods of output correspond- 
ing to input. The typewritten page is prepared by the console type- 
ivriter at 10 characters per second. Using this device, information 
about operation of stored programs, memory contents, and so on 
can be printed under program control or manually from the con- 
sole. 

Punched paper tape can also be an output device, operating at 60 
characters per second. This tape can be used in conjunction with 
the Flexowriter for the tabulation of results of operations. It can 
also be introduced to the IBM tape-to-card converter to produce 
punched cards for storage of data or further processing. 

Punched cards are, of course, a primary output of this or any 
system. Two card punches are available, either of which will operate 
on-line or off-line. Rates of operation are 100 or 250 cards per min- 
ute. In these devices, 10 computer words become the contents of 



132 Evaluation of equipment systems 

one punched card. Checking devices such as double punch or blank 
column detection are available. 

The "800" system Printer-Punch provides an output of punched 
cards and/or line printing. This device will produce 150 punched 
cards per minute or print at the rate of 150 lines per minute. 

Printed reports can be produced by two types of printers. The 
first of these operates at the rate of 150 lines per minute. The second 
operates at high speed, 600-900 lines per minute. Either printer 
can be hooked up off-line or on-line. Line length is 120 characters 
at 10 per inch. Vertical spacing is 6 lines per inch. Available charac- 
ters include 10 digits, 26 letters, and 20 special symbols. Vertical 
format is under program control. 

IBM 305 RAMAC 6 

General characteristics and information 

This is a general purpose computer, but is best applied to clerical 
and accounting applications where computation is at a minimum 
and file updating is the major system problem. This is a medium 
scale data processor which leases for $3,200 per month and sells for 
$189,950; these prices do not include peripheral equipment such as 
key punches, key verifiers, sorters, collators, and so on, if such 
equipment is required in the system. The "305" is a stored program 
machine which is generally set up to carry files in its disk storage and 
instructions in its drum storage. Control panels handle logical 
operations, format control, and input-output. 

Ramac is what is called an on-line machine. This means it is only 
limited by input and output rates and is not limited to any one 
sequence or volume of transactions. Remote inquiry stations enable 
company managers directly to interrogate the computer, with a 
monitor, at any time. This optional feature has many advantages 
since it brings the decision making process right up to the next in- 
line subsystem without human interference. On-line equipment has 
the ability to provide almost instantaneous access to voluminous 
records at any time the central computer is in operation. 

A normal "305" will require 15.1 KVA. This type of installation 

6 Manufactured by International Business Machines Corporation. 



Evaluation of equipment systems 133 

will require about 600 square feet (not including peripheral equip- 
ment or personnel) and will add a floor load of 8,925 pounds. Heat 
generation is negligible and normal air conditioning should be 
supplemented by air filtration and control over relative humidity 
(not to exceed 80 per cent). 

Programming and operator training are provided by the manu- 
facturer free of charge. Programming assistance, education, and 
library facilities are made available to customers. Data processing 
services are furnished on a contract basis to industry, science, and 
government by the Service Bureau Corporation, a wholly owned 
but independently operated subsidiary of IBM formed on January 
1, 1957. There are 82 branch office locations in principal U.S. cities, 
each operating punched card accounting machine systems. Seven- 
teen of these service bureaus are also equipped with electronic data 
processing systems. 

Number system 

The "305" is a decimal machine using binary code to represent 
decimals internally, with a parity bit. As in the previous computer, 
the "305" will handle digits through 9, alpha characters A through 
Z, and a variety of special characters. 

Instruction system 

Word length in the "305" is variable, although it consists of 10 
digits plus sign as in the previous computer. This is what is called a 
two-address machine because two addresses can be placed in the 
space of 10 digits. A picture of the "305" word (Figure 7-7) shows 
how the two address feature is achieved. There are about 15 basic 
instructions. The panel exit character stops successive program steps 



FROM 


TO 


Number of 


Panel 
Exit 


Special 
Control 


Track 


Position 


Track 


Position 


Characters 


Alpha- 
meric 


Numeric 


Numeric 


Alpha- 
meric 


Numeric 


Numeric 


Numeric 


Numeric 


Alpha- 
meric 


Numeric 





1 


2 


3 


4 


5 


6 


7 


8 


9 



Fig. 7-7, 



134 Evaluation of equipment systems 

and activates a panel hub corresponding to that character for opera- 
tion of selectors, input-output equipment, and so on. The special 
control digit can cause comparison of contents of from and to 
addresses and can setup panel connections, or it can cause the clear- 
ing of accumulators before entry. The control panel is used to per- 
form logic operations. 

Operations of the "305" are checked by reading twice and com- 
paring results. The normal parity checks are made to and from the 
drum, on the program register, printer, and punch. Addressing is 
an important issue with the "305" since five digits is optimum and 
any other arrangement of digits wastes space. There are many ways 
in which part numbers, stock numbers, and so on can be compressed 
to fit this five digit requirement and it seldom creates a problem. 
Time sharing on the "305" will permit any or all of these operations 
to be going on at the same time; line printing, card punching, seek- 
ing a disk record, card reading, and performing arithmetic or logic 
operations. Internal checks functioning on these operations are 
file write check (reads disk record for comparison with drum track 
after disk write), clock check (autocorrelates timing pulses), print 
check (setup on stick printer compared with characters on drum 
track), and double punch and/or blank column check (if wired in 
IBM 323 punch). 

Storage 

The "305" is unique in that its main storage consists of fifty disks 
which revolve at 1,200 rpm. Each disk has 100 tracks, and in the 
standard configuration each track contains 10-100 character records. 
Thus each "juke box" will hold 5,000,000 characters, and in special 
configurations contains 10,000,000 characters. This is a serial ma- 
chine. Average access time per record through the single arm is as 
follows: 



Time (milliseconds) 
Disk-to-drum Drum-to-Disk 

Minimum 30 80 

Average 55 1 05 

Maximum 80 130 



Evaluation of equipment systems 135 

New configurations with two access arms increase the amount of 
data which can be retrieved by 100 per cent, through proper pro- 
gram control. It is thus apparent that large amounts of data can be 
stored and retrieved very rapidly, even though interrogation to any 
one file in the disk unit may have a low frequency and occur in a 
completely random fashion. 

The instructions required co operate the "305" are carried in a 
separate magnetic drum which also acts as a buffer for input and 
output. One drum per "305" system contains 240, 10 digit words 
with a maximum access time of 10 ms. The drum, rotating at 6,000 
rpm requires a full revolution for decoding, transfer from drum-to- 
core, and transfer from core-to-drum, for a total of 30 ms to execute a 
command. The core storage, in this case, is a buffer and is address- 
able; characters can be transferred from but not to the core buffer. 

The "305" system is also available in a special configuration with 
the IBM 650 computer, with or without magnetic tape carriers. 
Prices quoted do not include these additional pieces of hardware; 
however, they deserve mention in this section since they enhance 
the computing capabilities of the "305" system considerably. 

The basic rate of the arithmetic unit is 100 records per second. 
The following are average times to obtain instructions and operate 
from storage: 

Operation Time (milliseconds) 

Addition or subtraction 30 

Multiplication 20 plus 10 ms per multiplier digit 

Division 20 plus 20 ms per quotient digit 

Transfer 30 

Branch 50 (via control panel) 

Shift 30 



Input methods 

Punched cards are the principle input medium via the card 
reader, one of which can be connected per system. Card read rate is 
125 per minute. Approximately 480 ms are required to complete 
input, but, through time sharing, 430 ms are available for other 
operations during this process. 

An auxiliary method of input is provided in the Punched Paper 



136 Evaluation of equipment systems 

Tape Reader, IBM 382, which will move data at the rate of 20 
characters per second. Input is also achieved through the console 
typewriter. 

Output methods 

Major output device is the punched card. Cards can be punched 
by the IBM 323 at a maximum rate of 100 cards per minute. Total 
time for output is 400 ms plus 20 ms per character, however, time 
sharing permits the use of the central processor about 350 ms plus 20 
ms per character, during this operation. One unit per system. 

The console typewriter can print 100 characters per line at the 
rate of 10 characters per second. One can be connected per system 
and is used primarily by the computer operator for debugging and 
error detection. 

The IBM 370 Printer can print an 80 character line at the rate of 
30 lines per minute. Because of its unique engineering features, on 
shorter lines of 20 characters it can print at the rate of 80 lines per 
minute. Printing is at the density of 10 characters per inch. Format 
is controlled by control panel wiring while the carriage tape con- 
trols feeding, spacing, and skipping with continuous forms. The 
"stick" printer can prepare up to eight good carbons. Frequently 
the IBM 402 or 407 printers are set up in conjunction with the 
"305" system to provide faster and more versatile output. 

UNIVAC SOLID-STATE 7 

General characteristics and information 

This machine is among the newest of the computers, and as its 
name Solid-State indicates, it is one of three all transistorized 
medium scale machines commercially available. In this same cate- 
gory are the RCA 8 501 system variants which fall into medium, 
intermediate, and large scale categories. The medium scale version 
of the "501" is called the RCA 502; the intermediate version is 

" Manufactured by Remington Rand Division, Sperry Rand Corporation. Carried 
in official bulletins as the "New UNIVAC Magnetic Amplifier Solid State Com- 
puter." 

8 The manufacturer of these systems is the Electronic Data Processing Division, 
Radio Corporation of America. 



Evaluation of equipment systems 137 

called the RCA 503; large scale is RCA 504. The IBM 1401 is 
similar in many of its design features to the "Solid-State," employ- 
ing high speed input, output, and processing features at very low 



cost 



The "Solid-State" computer is a general purpose system equally 
suited to commercial or scientific-engineering applications. It is a 
stored program machine utilizing both on-line and off-line equip- 
ment. An important feature of the "Solid-State" is its ability to 
accept data from both IBM and UNIVAC punched cards, unlike 
some of the other commercial computers. The "Solid-State" system 
leases for $6,950 per month and sells for $350,000. To this must be 
added the cost of input-output devices such as key punches, key 
verifiers, and so on, as required by data processing systems. 

The one-run approach to problems without tape units is a feature 
of this system. A typical one-run problem would be payroll or labor 
distribution where the introduction of all inputs would result in 
the completion of the run, uninterrupted by any need to divide ap- 
plications into parts, utilize multi-runs or use numerous machines 
to complete the task. Not all tasks can be so easily handled, but one 
pass processing, even in limited areas, recognizes a complex data 
processing problem, heretofore not attacked through equipment 
design. 

The "Solid-State" functions without control panels of any kind. 
Because there are only fifteen vacuum tubes in the entire system, 
the power requirement is only 15 KVA. The system weight is 6,000 
pounds and the space requirement is about 400 square feet, not in- 
cluding peripheral equipment or personnel. No air conditioning is 
required. 

Formal training courses are provided free of charge by the manu- 
facturer. Representatives will spend several months with the cus- 
tomer to assist in training personnel and setting up a working organ- 
ization. The manufacturer provides programming assistance and 
maintains a central exchange facility for distribution of library 
routines. A programming and coding staff is available on a contract 
basis for special problems. Computer facilities are available from 
the manufacturer on an hourly basis or on a fixed fee basis for a 
specific problem. These facilities are also available prior to the in- 



138 Evaluation of equipment systems 

stallation of a customer's machine for checking of preparatory cod- 
ing and for running sample problems. 

Number system 

This computer utilizes a four bit biquinary type of binary coded 
decimal system, plus a parity bit, which is automatically checked 
moving to and from the drum. In its internal operations, the "Solid- 
State" handles all character bits simultaneously, which increases its 
operating speed. 

Instruction system 

The "Solid-State" word is ten digits plus sign and space. The 
50,000 storage locations are arranged in groups of 200 words to 
make up 25 bands. Average access to normal bands is 1.7 milli- 
seconds; access to fast bands is .425 milliseconds. Figure 7-8 is a 



OPERATION 
CODE 

1 I 2 



OPERAND ADDRESS 
3 | 4 |5 |6 



NEXT INSTRUCTION 
ADDRESS 

7 | 8 | 9 | 



SIGN 



Fig. 7-8. 

picture of a "Solid-State" word. Normally, instructions go through 
a four phase cycle: (1) obtain instruction, (2) prepare to operate on 
instruction, (3) search for the operand, and (4) execute the instruc- 
tion. The previous instruction directs the machine to perform its 
next operation on the current instruction. This is known asaii^ 
address system. A cycling unit generates timing signals and, in con- 
junction with a timing circuit, detects any lack of drum synchroniza- 
tion and automatically signals this condition. 



Storage 

Data and instructions are stored on a magnetic drum. The "Solid- 
State" stores 5,000, 10 digit words on its 25 bands. Drum speed is 
17,500 revolutions per minute. There are four read-write heads on 
fast access bands which reduces access time to a very low point. Time 



Evaluation of equipment systems 139 

to extract a word from the drum under these conditions is 17 micro- 
seconds. Three buffers are utilized to synchronize input-output 
transfer with the higher speeds of the central processor. Drum 
buffers communicate directly with the intermediate storage buffers 
which are housed in input-output devices. The capacities of these 
buffers are in addition to the 5,000 words of the central drum. The 
following capacities are, thus, added to the system: 





Characters 


Print buffer 

High speed reader buffer 

Read-Punch buffer 


130 

200 
200 




530 



Data transferred to the input-output buffer areas may be punched 
or printed independent of the central computer's operation. As a 
result, time sharing is one of the important features of this system. 
More complex configurations of this computer are offered in con- 
junction with UNIVAC magnetic tapes (UNISERVOS). Thus, ex- 
ternal memory will be available to complement internal storage. 
The following are times for the basic arithmetic and logical opera- 
tions: 



Operation Time (milliseconds) 

Addition or subtraction .085 

Multiplication .255 to 1.79 

Division .425 to 1 .96 

Branch (Compare) .051 

Transfer .068 



Input methods 

Two basic input methods are currently available. The first is 
via punched cards into the High Speed Reader. This machine 
operates on input at 950 cards per minute, accepting 80 or 90 col- 
umn cards. This reader has a selector device that allows the feed of 
cards from any one of three hoppers into the card reader, each 
hopper containing about 1,000 cards. Double reading of input data 
provides a check on input validity. Cards are selected for reading 



140 Evaluation of equipment systems 

from stackers, and are stacked in hoppers under program control; 
output hoppers have a 1,200 card capacity, each. Thus, the reader 
has actually more than the normal input function; in addition, it 
can collate data in the system and also segregate input data. The 
cycle for this machine is 133 ms, however, time sharing makes avail- 
able all but .51 ms, because of buffering. 

The second method of input is through the Read-Punch Unit. 
Operating at a maximum rate of 150 cards per minute, it can func- 
tion either independently or in conjunction with the High Speed 
Reader. Like the former device, the Read-Punch Unit has the func- 
tion of entering and/or collating data into the system. It has the out- 
put function of punching cards as a result of processing. It has the 
ability to read a card and punch computed results into the same 
card. It can segregate output cards and perform the same audit as 
is done in the High Speed Reader. Time sharing ties up the central 
processor only 6.82 ms per operation, which leaves ample time for 
audit and checking, stacker selection, computing, simultaneous 
operation with High Speed Reader, sensing, and punching. 

Output methods 

The most important output method is the High Speed Printer. 
which operates at the rate of 600 lines per minute. Control is by 
means of stored program. There are 130 printing positions possi- 
ble, per line, each position being capable of printing 26 alpha, 10 
numeric and 15 special characters. Characters are spaced 10 to the 
inch; line spacing is 6 to the inch with single, double, and triple 
spacing optional to the programmer. Report format, zero suppres- 
sion, and so on are also under program control. Paper is 4 to 21 
inches wide. An original and five carbons are possible at these 
speeds. 

SUMMARY AND CONCLUSIONS 

There are a large number of additional computer systems which 
have not been mentioned. IBM has its well-known, medium scale 
650, its new, intermediate scale, solid state 7070, and its large scale 
705 (I, II and III), 709 and 7090. Remington Rand has its Series 



Evaluation of equipment systems 141 

60 and 120 small scale equipment as well as the File Computer, 
UNIVAC Scientific (1 103, 1 103A) and UNIVAC II. National Cash 
Register and other companies offer equipment very worthy of con- 
sideration. Each of these systems is a major contribution to the 
science of data processing; each possesses individual, significant 
characteristics which may be put to the advantage of a potential 
user. A technically oriented equipment evaluation would maximize 
the likelihood that all equipment which qualifies for the task would 
receive objective and adequate consideration. 

The task of the systems engineer must begin to emerge in this 
new dimension. Who is to select the equipment complement best 
suited to the data processing tasks? Certainly, this can't be at- 
tempted until the data processing tasks are not only known, but 
organized into a data processing system. The accent here is on the 
use of systems analysis to design business systems compatible with 
each other, with clear recognition of the organization's goals. 

Equipment evaluation must be done by experts. Experts may be 
systems engineers or they may be specialists in data processing 
equipment. A knowledge of machine programming is required to 
evaluate effectively. A knowledge of hardware is also desirable in 
this work. Most of all, a knowledge of systems goals must be used as 
the overriding criterion in determining how the attributes of each 
computer system can be put to best use in the company's data proc- 
essing. Where the computer becomes the data processing device, the 
systems analyst is an essential ingredient. Efficient systems design 
may make the difference between a loss, break-even, or profitable 
system operation. 

It must also be obvious that data processing is a restrictive device 
in that it does not permit a wide variation from rules or established 
patterns. Again, the role of the systems analyst becomes critical, for 
it is he who must define the system boundaries and then the sys- 
tems, in complete detail. The job of describing system elements, 
system requirements, and system input-output can only fall to 
specialists who have the training and ability to conduct methodical 
analysis in a field which resists discipline and improvement. 






8 



Constructing 
the system costs 



WHAT ARE SAVINGS? 

Savings is a word about which there is little semantic difficulty in 
everyday life. In the business world, its definition of economical or 
not wasteful falls far short of the need. Typically, "savings" will be 
claimed when a cost advantage of any kind exists, but far from 
being economical, such an achievement might require the expendi- 
ture of many dollars, before the economy can be realized. Not waste- 
ful is an equally inadequate definition: Looking at the burn-off of a 
typical oil refinery which may light up the sky for miles around it, 
one can only conclude that it must be less expensive to waste this 
potentially useful commodity than to reprocess it in some marginal 
use. 

Cost advantages equal savings. Is this true? 

There are undoubtedly many, many circumstances when it is, 
but in the design of systems, it can frequently be untrue. First, we 
must determine what a cost advantage is. One definition might be 
something which ultimately reduces the expenditure of funds. 
There are basically four categories of expenditure: labor, material, 
services, and equipment or facilities. Thus, we are saying in effect 
that a true savings is one which will either reduce or eliminate the 
current costs which are being incurred in one of these four cate- 
gories. Now let us test this hypothesis. 

An accounts payable system is designed which will require the 
addition of new equipment. No personnel will be released; some 
forms will be scrapped; slightly larger facilities will be required. 
What are the advantages? 

142 



Construe ting the system costs 143 

(1) Savings in time to post each account payable 

(2) Reduction of errors in posting 

(3) Faster reporting of accounts payable balances 

(4) Analysis of accounts payable data for the benefit of Accounting 
and Purchasing 

(5) System capability to absorb increased work load of 33 per cent 

Would you approve the installation of this system on the face of 
these facts alone? If the new equipment and forms were budgeted at 
$5,000 per year, would you then accept the proposal? What if the 
additional costs were $20,000 per year? 

This is a typical no-relief situation. That is, there are no real off- 
setting dollars to balance against the costs to be incurred by estab- 
lishing a more desirable accounts payable operation. There is some 
possibility that item 5 might provide ultimate dollar benefits: 

(1) If new activities that would require the hiring of personnel are 
transferred to the accounts payable group, thus eliminating the 
need to hire; or, 

(2) If existing activities can be merged into the accounts payable 
group, eliminating marginal personnel, equipment or facilities 

Against this no-relief situation, must be balanced the desires of 
Management, who as systems purchasers have their own set of rules 
and values. If this appears to be the only way to eliminate human 
error, Management might be moved to act. Likewise, if Manage- 
ment anticipates increasing growth, the capability represented by 
a man-machine system might make this an attractive proposition. 
Some companies might accept the increased costs of such an im- 
proved system, merely because they feel it is in their best interest to 
employ the most modern accounting methods which are available. 
A printing establishment was approached by a consultant who 
studied his potential client's operation for a week, and devised this 
four point program: 



Area to be studied Reason for study 

1 . Redesign of incoming order data proc- 1 . Reduction in order processing time 
essing system. will provide a competitive advantage 

by enabling your firm to fill orders 
sooner. 



144 Constructing the system costs 

2. Re-establishment of composing room 2. Better standards will enable Produc- 
and print room standards. tion Control to balance work flow be- 
tween composing and print, which 
will decrease the time some orders 
will spend in-process. Better standards 
will have a tendency to increase pro- 
duction. 

3. Re-establishment of pricing policies 3. Better standards will provide a more 
based upon better standards. accurate relationship between costs 

and selling price; more accurate cost 
data will give more flexibility in bid- 
ding; more accurate cost data will give 
the opportunity to improve competi- 
tive position in jobs where a higher 
number of bids are rejected. 

4. Re-allocation of salesmen's time. 4. Statistical analysis will provide more 

data on how to use salesmen effec- 
tively. 

None of the reasons given for working on the client's problems can 
be termed savings. A better word might be benefits. But the payoff 
in each case is not clearly related to dollars. If we assume that by 
processing orders in some unique way, two days in-house time can 
be saved, does this really mean the client will do more business? 
And if it is true, how can the dollar return to the client be identified? 

Although better standards may have a tendency to pace operators, 
how can the increased rate be measured in a printing establishment 
that operates like a job shop? The mix of orders could be deceptive 
and cause the flow of work to move faster or slower. Balancing pro- 
duction is desirable, but, again, if we are going to accept or reject 
the consultant's proposal on the basis of dollars saved or increased 
earnings, we must be able to measure a very tricky, elusive value 
since the over-all status of each of our criteria may be affected by a 
number of factors simultaneously, over a long period of time. The 
inability of the existing accounting system to pick up the requisite 
detail could alone frustrate this proposal. 

Item 3 is potentially an excellent project. If this portion of the 
presentation was backed up by actual data on completed jobs and 
rejected jobs, this Management might have been very enthused 
over the possibilities. Here it would have been important to state 
exactly what new tools would be used in refining the existing system 
— and perhaps some indication of the way in which the new system 
would operate and the actual dollars it might be expected to return. 
It is no fallacy to say that, in the consulting business, the problem 



Constructing the system costs 145 

must be "half-solved" before one begins to work. One might ques- 
tion whether this consultant could prepare for such a project and 
the other projects he has suggested in only a week. 

The salesmen project, as stated, is very weak. Although it is possi- 
ble to analyze numbers of calls and numbers of sales in various ways, 
the payoff to Management is not very clear. If ten calls per day pro- 
vides three sales, and Management wishes to increase sales, might it 
not be a lot simpler just to have more salespeople? The alternative 
of telling a salesman to stop calling back after the second, third, or 
fourth call, is not a very powerful incentive in terms of the dollars 
Management must expend on consultant's fee. 

As long as profit is the Management motive, savings or benefits 
will almost invariably have to be tied to the reduction or elimina- 
tion of existing costs. There are a number of exceptions where Man- 
agement must create a new capability, and in this case the motive 
may be to achieve a specific level at a minimum cost. Here, it may be 
necessary to make a trade-off and have slightly less capability because 
of the reduced out of pocket expenditures. If maximum capability 
is required, cost may be no factor. In many instances, time will be 
substituted for cost in this equation. For instance, DEW (Distant 
Early Warning) line radar capabilities were justified on the basis of 
time saved in data transmission rather than in terms of cost. Many 
of our current military projects represent such capabilities which 
have been deemed essential under current international conditions. 

For the systems analyst, the systems purchaser is a primary factor 
to consider. Although it may be necessary ultimately to change or 
alter the purchaser's original view of the system (if this is possible), 
the objectives which are initially established will have a great bear- 
ing on the ability to demonstrate savings or benefits. The ability to 
demonstrate actual dollar savings is the most devastating test of a 
system which has been avowedly designed with this as the sole 
criterion. 

Intangible savings, however, are equally important. They are 
called intangible not because they do not exist, but because it is 
difficult to ascribe a dollar value to the extent of the savings. How 
does one put a price on faster reporting? Or how would you evaluate 
the future expandability of a system? Of course, it is possible to 
place values after each of these statements and attempt to justify 



146 Constructing the system costs 

their tangibility. Sometimes, however, there is a high cost attached 
to this type of activity. Finally, Management is usually broad- 
minded enough to recognize the futility of trying to exhume the last 
dollar of accuracy in a proposal, and will look at the easily computed 
savings, if such exist. The proposal to Management cannot rely on 
these intangible factors, but they should and must be included. 

ONE-TIME VERSUS RECURRING COSTS 

Proposals of cost and savings require a statement of both one- 
time and recurring costs (see Table 8-1). One-time costs are those 







TABLE 


8-1 








Electron 


ic Data Processing System 






Comparison of 


One-Time 


and Recurring Costs 








system I 


SYSTEM II* 


One-Time Costs 




Rental 


Purchase 


Rental 


Purchase 


Prepare facilities 




$ 13,000 


$ 13,000 


S 15,600 


S 15,600 


Systems studies 




12,000 


12,000 


12,000 


12,000 


System conversion 




8,600 


8,600 


8,600 


8,600 


Parallel operation 




11,000 


11,000 


10,800 


10,800 


Miscellaneous 




4,000 


4,000 


2,000 


2,000 


Total one-time costs 




$ 48,600 


S 48,600 


S 49,000 


S 49,000 


Recurring costs 




Rental 


Purchase** 


Rental 


Purchase** 


Equipment: 












Electronic System 




$101,100 


$127,000 


S 51,800 


S 34,900 


Tabulating System 




80,000 


80,000 


72,000 


72,000 


Maintenance 




(included) 


63,000 


(included) 


68,300 


Personnel: 












Operations 




30,100 


30,100 


28,300 


28,300 


Applications 




26,400 


26,400 


26,400 


26,400 


Systems Analysts 




24,500 


24,500 


24,500 


24.500 


Supplies 




20,000 


20,000 


20,000 


20.000 


Total recurring costs 




$282,100 


$371,000 


S223.000 


$274,400 


Total annual five year 


costs 


$330,700 


$419,600 


$272,000 


$323,400 



* Equipment cost prorated over that part of the first shift actually utilized (about 
40%). System I is at 80% of first shift utilization. 
** Purchase price prorated over five years. 

for the establishment of the new system. Recurring costs are those 
which will be sustained to operate the system after it has been estab- 
lished. In the establishment of a proposed electronic data process- 
ing facility, the costs in Table 8-1 were evaluated for two computer 



Constructing the system costs 147 

systems which had approximately the same data processing capabil- 
ity. It is proper to prorate equipment purchase costs over a five year 
period if this is what Internal Revenue will permit. However, one 
might challenge the logic of only admitting 40 per cent of the costs 
of System II, which makes it appear as if this is the cheaper of the 
two systems. 

There is no yardstick to say how much the one-time costs of a 
system may be, since the systems requirements of each problem will 
determine how long the investigation and hypothesis must be. The 
Work Assignment form (Figure 5-1), attempts to correlate as much 
of this type of data as possible. Clearly, however, the one-time costs 
are relatively unimportant, if there are real savings to be obtained. 
Eventually, the amortization (spreading of costs over some period of 
time) of costs will end, but the systems savings will continue, unless 
subsequent decisions alter the systems design. 

It is important to include all relevant costs. Sometimes identical 
costs may be eliminated to concentrate attention on differences. 
Generally, this is not desirable because this has the effect of under- 
stating the present and proposed systems. Frequently, overhead is 
not included in such computations because of the attitude that fixed 
costs go on, despite systems design of one type or another. This may 
be true, but there are circumstances when it is not. One computer 
system might, for instance require no air conditioning; another 
might require humidity and temperature control. Sometimes, the 
decision to take on additional overhead will affect variable costs. If 
System A requires 40 people and System B requires 13, there are 
additional costs to be borne by System A to pay for the indirect costs 
of personnel. This was precisely the case in a timekeeping system 
which promised Management different levels of accuracy in direct 
labor reporting. It was the indirect costs of personnel that forced the 
selection of System B. 

There is sometimes a tendency to misstate recurring costs, either 
through a desire to sell a good idea or because the system problem 
cannot be precisely appraised. In either case, Management is seldom 
sold on either side of a marginal cost or savings argument. If the 
margins aren't convincing, Management will tend to look more 
closely at the benefits and "intangibles." Precise appraisal of the 
load the system must carry and the allowance of adequate safety 



1-/8 Constructing the system costs 

margins may make the picture less enticing, but they invariably 
leave the systems designer in a less vulnerable position. 

DISPLACEABLE AND NON-DISPLACEABLE COSTS 

If possible, the costs of the analysis should be related to the results 
achievable. For instance, it might actually require a minimum time 
of two weeks to set up a system of accounts receivable for an insur- 
ance agency. If the costs of this system design and installation are 
set at $4,750, it is advisable to determine immediately whether or 
not there are $4,750 of offsetting costs to make the project self- 
liquidating for the systems purchaser. Offsetting costs are either 
non-displaceable (they remain after the new system is in and operat- 
ing) or displaceable (they are removed with the full implementation 
of the new system). In the example cited above, the time of 1.25 
clerks was involved in accounts receivable; this was equal to S500. 
per month on this company's pay scale. The project was temporarily 
put aside because the dollars displaced were not sufficient to liqui- 
date the cost, in a reasonable time period. Table 8-2 compares the 
two systems: 

TABLE 8-2 

Cost to design new system S4750.00 

Monthly cost to operate 

present system: 

Labor $500.00 

Facilities 100.00 

Supplies 40.00 

$640.00 

Monthly Cost to operate 
new system: 

Labor S200.00 

Facilities 1 00.00 

Supplies 60.00 

Service Costs 60.00 

$420.00 

Available to amortize 

one-time cost $ 220.00 

Months to pay off investment 21 .6 

This example has very complex counterparts. For example, in the 
proposed introduction of an electronic data processing system, the 
first step would be to create an accurate breakdown of existing 
costs. Personnel would be analyzed by job classification. Equipment 



Constructing the system costs 149 

would be listed in complete detail with a system of references to tie 
personnel to equipment (it is essential to know which clerks operate 
the flex-o- writer, electric typewriter, comptometer, and so on. From 
lists of this type, certain personnel and equipment would be elimi- 
nated (displaced) because new equipment and /or personnel would 
be added back. The figures which describe the net existing systems 
costs are the non-displaceable costs; that is, they cannot be elimi- 
nated and will be combined with certain additional costs. The non- 
displaceable costs plus the added cost figures are termed the new 
systems costs. 

Table 8-4 shows a summary of costs in the displaceable and non- 
displaceable categories. The annualized non-displaceable costs are 
now ready to be added to the annualized one-time and recurring 
costs. The summary in Table 8-3 has been drawn from data in 
Tables 8-1 and 8-4. Table 8-4 is the chart, if the full costs of System 

TABLE 8-3 



SYSTEM I 

Rental Purchase 


SYSTEM II 

Rental Purchase 


Total annual 
five-year costs 

Total non-displace- 
able costs 

New system costs 

Old system costs 
per year 


$ 330,700 

$ 969,128 
$1,299,828 

$1,512,288 


$ 419,600 

$ 969,128 
$1,388,728 

$1,512,288 


$ 272,000 

$ 969,128 
$1,241,128 

$1,512,288 


$ 323,400 

$ 969,128 
$1,292,528 

$1,512,288 


New system 

savings per year 


$ 212,460 


$ 123,560 


$ 271,160 


$ 219,760 



II are charged against the 40 per cent of utilization on which previ- 
ous figures have been prorated. 

TABLE 8-4 

Electronic Data Processing System 

Schedule of Displaceable and 

non -displaceable costs on an annual basis 



PRESENT SYSTEM COSTS: 






Production Control 






Personnel 


( 85) 


$595,000 


Equipment 


( 8) 


100 


Space 


(3500) 


350 


Tabulating 






Personnel 


( 40) 


$280,000 


Equipment 


( 21) 


201,500 


Space 


(2000) 


200 



150 Constructing the system costs 



TABLE 8-4 (Continued) 



PRESENT SYSTEM COSTS: 










Accounting 










Personnel 


( 28) 




5196,000 




Equipment 


( 14) 




800 




Space 


(2000) 




200 




Engineering 










Personnel 


( 14) 




8126,000 




Equipment 


( 1) 




18 




Space 


(1200) 




120 




Other Personnel 


( 16) 




SI 12,000 




TOTAL PRESENT SYSTEM 


COSTS 








positions eliminated: 










Production Control 


( 24) 




SI 68,000 




Tabulating 


( 2) 




14,000 




Accounting 


( 4) 




28,000 




Engineering 


( 4) 




36,000 




TOTAL POSITIONS ELIMINATED 


( 34) 




S246.000 


EQUIPMENT ELIMINATED 










Production Control 


( 4) 




$ 50 




Tabulating 


( 14) 




184,340 




Accounting 


( 10) 

INATED 


( 28) 


580 




TOTAL EQUIPMENT ELIM 




SI 84,970 


SPACE reduction: 










Production Control 


(1000) 




$ 100 




Accounting 


( 500) 




50 




Engineering 


( 400) 


(1900) 


40 




TOTAL SPACE REDUCTIOI 




S 190 


OTHER PERSONNEL ELIMINATED 


( 16) 




SI 12,000 



TOTAL DISPLACEABLE COSTS 
TOTAL NON -DISPLACEABLE COSTS 



SI, 512,288 



S 543.160 
S 969.128 



TABLE 8-5 





SYSTEM I 


SYSTEM II 




Rental 


Purchase 


Rental 


Purchase 


Total annual 










five-year costs 


$ 330,700 


$ 419,600 


$ 606.500 


S 735,000 


Total non-displace- 










able costs 


$ 969,128 


$ 969,128 


S 969,128 


S 969.128 


New system costs 
per year 


$1,299,828 


$1,388,728 


$1,575,628 


SI, 704. 128 


Old system costs 
per year 


$1,512,288 


$1,512,288 


$1,512,288 


SI. 5 12. 288 


New system 

savings per year 


$ 212,460 


$ 123,560 


S* (63,340) 


$*(191,840) 



* Figures in parentheses are additional costs. 

Now System II costs no longer look as attractive as before, at least for 
the first five years. However, a decision to order System I could be 



Constructing the system costs 151 

shortsighted. System I is at 80 per cent of utilization, which means 
any additional heavy data processing burdens will push into a 
second shift. Second shift rentals are charged at 40 per cent of the 
first shift rates, for computer equipment time actually utilized. 
This, of course, tends to reduce the average hourly cost of equip- 
ment. Against these savings, the inconvenience of second shift opera- 
tion must be weighed. If equipment is to be purchased, the second 
shift operation merely means that amortization will be spread over 
that many more hours of use, so actual operating costs will not be 
reduced until the system is paid out. 

The amount of additional data processing hours to be connected 
will have an important bearing on whether the choice will be Sys- 
tem I or System II. There are some time considerations, as well: It 
may be desirable to select System I on a short term basis, and repro- 
gram (a semi-automatic operation) for System II at a later date. 
Thus, another variable of time of conversion has been injected, and 
should be evaluated for consideration. If conversion will take two 
years, System I savings will be enjoyed for some period before second 
shift operation is required. At a certain point, second shift opera- 
tion may be undesirable, and then System II installation can be 
reconsidered. 

Growth or decline of the existing system could be an important 
factor, as well. If the costs of the present system expand in direct 
proportion to the number of transactions, file size, or some other 
factor, it may be appropriate to inject anticipated growth into the 
dollar figures, to look ahead to problems which are not currently 
pressing. This is desirable when it can be done with some accuracy. 
Long term committments require this accuracy, since Manage- 
ment may wish to make a decision in favor of the long range policy, 
which extends the payoff into the future. Examples of such projec- 
tions are illustrated in Tables 8-6 and 8-7 for an expanding elec- 
tronic data processing center which will need to provide different 
levels of service over a three year period. 

Side benefits may be added to the dollar pool: For instance, there 
may be, as a result of a new system, generally increased labor utiliza- 
tion. This might be expressed through the decrease in waiting time 
at tool cribs or reduced waiting time for new job assignment. Per- 
haps a system will carry along with it the benefit of reduced inven- 



CM 










CO 


co CO O 


tH WO 


-H ^H O 


*n 


N 


CO 1*1 


^r o 


o © 


-r 


i-t co 


CM O 


FH CO 


cm 


*y 


to" CM 


CM 






CM 










to 











O O -h o t^ h 00 (M 

CO -^ co co (M C 
CO © © X — 



I* © m i - 



r- co o t- CM O co 

Tf -<f 1TJ © © 

© CO O O 

iO CM CM 



O l> 

© co 
CO CM 



O CO 
co co 
co co 



(C — i^ co 

© CO — CM 

— © co i^- 



© © © OCMO tH-^O .— fhO 

00 tj* tH CO O © m CO 

i>« co a\ m co —h co 

io Of m 



Nrtlfl 


r^ i— i oo 


CM — O 


fM 


O CO 


© CM 


O »0 




m Si 


oc 


_ i rfi 


t^ 



CO 

-~~ - 00 co O 

CO co "f 



I-J — 



HNO tjh m O ■* 
O (O Ci OS TjH 

© lO CO —i 



O m <— i xo t^ — i 00 CM — O co — 00 

co 00 CO CO CM © >-0 00 ^f 

co ^f 05 00 —i rr co t> 

V — 



l> iO 03 CO 



© m © 

lO >-H 



CM CM © CO — i © © 
CO © © © Tfi 

00 iO co —i 



© 


t^ H lO 


h« 


CO 


r^ © 


© 


co 


rfi © 





© 
© 

~-- m m © 
© co m 

I GO © 



©CM© CM— i© CO — < © 
© © © © CO CO 

CO lO © —< CO 



© — i © CM >— < © -H 
rf J>- © © CO 

© CM © — 



F- 1 tfl *». 

© © 
© 



00 CM 
CM © 
X — 



© CM 
iO CO 
■<*" © 



© 


00 — < lO 


^rt 00 


CM - 


H © 


— 


oo 


co © 


© ox 




iO 


r-. 


CO 


<M © 


oc 


1 — 


■<*l 


CM 



m i> © t cm oc 

— CM — CM CO r- 
lO © lO CM CM 



©_ tt CM CM 2* 



© ~ 



Tfn CM © 
CO 00 



© — < © © 



iO 00 ph M ©-h© © 

© cm io r> 

© 00 rfi — 



cm r- — © x 

l". r f{ t^ 

— CM — t CM 

X X 



.2 » 

£ .o £ 

o£ o 

3 ° s 

o 6 o 





H Z h 



o o o 
HZ h 



o o o 



^ 



ffi^Q X^Q K^Q 

IH U-l u-i 

o o" . o odo oo'o 



q ^ c 



Q -3 Q S Q 8 n 



. o o o o 



Z r 



e 



cu 

§ s 



3 « S 



CM 


© 


m 


iO 


m 


© 



CM 
















CJ 


-f - 

lO 

© 


-i oo 
CM 
CO 


cm <-• or 
in oo 


■f O O 00 00 T* 
-h i^ 00 

rt" CO — ' 


CM CM © 

cm m 

m cm 


rjH ^ m 

CM CO 
CM 00 


© -4 00 
O CM 

m co 


1 




CO 


CM 


oo m 


© 


— | CO 


CO 



O-hO © in © © -if 00 -f 00 

r-~ t^Q m — i h o oo o 

oo ^ © oo^ -h co © p 

cm" in V ©" — <" -h" 

r— < m m 



or 




































CO 


r-H ,-H 


no 


~h CM 


— i 


o 


o 


m 


X 


■* 


m 


CM 


CO 


a 


CM 


o 


to 




CO 


CM 


CO 


00 


O 






8 




X 


o 




>o 


to 






to 


CM 

O 


© 


co 


x^. 


l^ 


<* 










oo 




CI 


o 




to 


CI 






































CO 




CM 








oo 




m 


l-H 




cr. 


CO 




r-» 


1 ' 



oo mcMO o m go h^ cm © o 

cm j>- rt<o m win - . o — i 

oo i-4 o p co in in r~ eo^ 

co" cm -f in tj? o" -h -h 

-h © to 

%<3= <=^= &^ 



^OJ~ 

i 



00 


TM — i CM 


CM 


© 00 


CO 


to r-~ 



COO© r—oOTfn -f CM © 00 CM O CM 

— i cm oo Tt< m Tf r^ — j 

CO X>; l-H © CM -T*i ©, © 

CM t»» m i-h © CM 1> i-h 



oo ocM© o m> © ^ ~f CM© © 

cm tjh -ho m cm m — i m i-h 

co^ j>« ©„ ©„ °o »o m t- go 

co" -h ^h in th' o" — a — r 

i-H CO CO 



CO GO 

~~~- 00 
© CO 



2 © 

I 

© 
© 



00 1> 

CM 00 

co m 

GO 



00 © 

CM O0 

oo^ in 
go" 



CM O©© © 00 tjh -h CM © CMCM© t- 

oo © go oo go mm i> © 

!>. GO © i— I i-H CM © © © 

cm i> xn i-h oT cm x> —T 



CM -hO© GOt-© OOCM© ©CM© 

oo© t^oo-^mmr^ 

i>. go tH cm cm cm oo © 

cm" i> oo h © cm i> 



oo mcMO om© h^ cm© o 

cmoo tjh © m cm m -h m — < 

m oq ©p oq_ in m t^ co_ 

oo" —T t}h m" th" o" i-h p-h 

^H © © 



00 CD (MO om© o© oo 

CMCO tHO in CM m -f CM 

go ©_ OO © CO ©^ tH 1> 

co" cm tjh m" -f" oo" i-h 

i-h m 



w a 

o o 
9 U 



3.3 



O fa 

5 ° 

W z 

o 



00 !>• -h CM CM©© CMl^ CO 't CM CD 00 -h m CM 

CMm x o co oo "f m-* 1 go -^ 

oo^ m t^ om ©^ cm © cm ^ oq q 

GO CM b CO h a n w h 



O CM O 
i>- © 


© m © 
© m 

© 00 


•f ^h 

-h CM 
00 


00 -H- 

© 00 
GO © 


m 


-H TH 


m ^ 


m 


■- 1 


m 
m 



X 

ci 
so 


-HO© ©©© CO ■* CM 
tH © "t 1 © 
go cm © m 


m © oo 
m © 
© r> 


© go m 

© —* m 

CM 00 


© m © 

00 rH 

tH co 


425 

10 

700 


© m © 
© m 

o 00 


tH cm 


CM -H 

CM -h 
© GO 


SO 


CO 


th r^ 


m r^> 

CM 


GO © 
i-H TjH 


m © 


© tH 
CM 


m ^t 1 




CM i-H 


© 

— 



© © © CM © © 
O 00 



ci 



CTJ 


m 


00 


© 


_H 


4< 


_ 


© 


to 


© 


00 


m 


m 


m 


GO 


C 


















m 


GO 


GO 


00 


T)H 

OO 


GO 


GO 


© 



CM © 



m 



OmO ^ CM CMO X 

o m co —i oo — i cm 

p_ cc -f cm — i m 

©" cm -h — " 

o o 



| 1 S 

O m o 

X * Q 





u 






OJ 






u 






q 












a 




■J~. 


c 




M 


C 




W5 


y 


5 




s 






s 


:- 




■=, 


^ 




CJ 


CJ 


n 





;j 





o 


u 


n 


ffi 


Oh 


Q 


X 


Ph 


Q 


I 


(0h 


Q 



CJ CJ CU CJ CJ 

c e c c c 

rjc/i C 5 ' 1 C?" C 175 flf 

111 111 !§! Ill I il 

Ocjo Ocjo Otuo Ocjo 4) o 



rt^rt a w cc rt " rt rt u d pj w ccj rt w cti rt^cti rt^aj a3 u rti 

ooo oo'o ooo ooo ooo odo ooo oo'o odo 
HZH HZH hZh HZh HZh HZh HZH hZH HZh 



. — cr a 

o o cj o 



e^ H 






.2 S 



(h 

en 



Sh O 

■i-i ro 

3 !h 

u 



bo fc 

c P 



a 



«j 



■So 

H 



^ 2 

3 cj 

Dh Dh 

cu w 



CJ 

3 w 

Oh -y. 

S -^ 
o si 

Sh 



4J u5 

If 

i< 

u 



H 13 



154 Constructing the system costs 

tory; in this instance, it is perfectly correct to add to the projected 
annual savings the dollars required to carry that portion of the in- 
ventory that will be eliminated. 



THE PRESENTATION OF SYSTEM COST DATA 



The fundamental rule is simplicity. Keep in mind the audience 
level and gear the facts accordingly. The audience at best can only 
take away one or two new ideas — so organize the presentation to be 
certain the most important points get across. One useful guide is to 
(1) tell them what you're going to tell them; (2) tell them; and then 
(3) tell them what you've told them. Repetition is generally a good 
idea to assist listeners in absorbing new ideas. Length of presenta- 
tions should be limited; almost any story can be told in an hour. If 
the time exceeds an hour, the audience's attention will begin to fall 
off rapidly. A question period after the presentation is generally 
useful because it brings the audience actively into the problem and 
may revive flagging interest. Color is a valuable aid in maintaining 
attention. 

Many a good system has failed in its presentation simply because 
of lack of organization. A general approach in preparing presenta- 
tions is to think in terms of outlines. If the story of the systems study 
will be presented on flip charts, the following outline suggests how 
to keep the package logical: 

Chart I: The Content of the Presentation 

A. Statement of the Problem 

B. Alternate Solutions to the Problem 

C. Recommended Solution to the Problem 

D. Benefits to Be Derived 

Chart II: Problem Statement 

A. Background Conditions and Method of Study 

B. The Existing System Operation 

C. Problems of the Existing System 

D. Areas Requiring Modification 

Chart III: Alternate Problem Solution A 
A. New System A 

(1) Advantages and Disadvantages 

(2) Dollars Saved 



Constructing the system costs 155 

Chart IV: Alternate Problem Solution B 
A. New System B 

(1) Advantages and Disadvantages 

(2) Dollars Saved 

Chart V: Selection of the Recommended System 
Chart VI: Benefits of the Selected System 

Note that Chart I tells the full scope of the presentation. Point LA 
is expanded in Chart II; point LB is expanded in Charts III and IV; 
point LC is expanded in Chart V; and point LD, which summarizes 
the presentation, is enlarged in Chart VI. Obviously, the design of 
the presentation must be dictated by the problem, and this sequence 
merely illustrates the principle of conciseness combined with a logi- 
cal approach. The spoken words which accompany the display of 
each chart are, of course, the most important factor in a successful 
presentation. Careful organization of significant points in outline 
form will help to keep the idea which is being developed clearly be- 
fore the audience. It is valuable to provide the audience with a 
take-away summary of the presentation which can be handed out 
before the question period begins. If the audience is given paper 
and pencils to make notes of questions that occur to them as the 
presentation is made, interruptions can be avoided. 

Since selling the idea can be a significant factor in the success of 
a systems program, time spent to prepare the presentation is as im- 
portant as any other phase of the program. A rule of thumb is a man- 
hour for every man-day; for instance, a task of ten man-days should 
require about ten man-hours for organization. 



9 



Operations research 
in business 



The purpose of this chapter is to provide a basic understanding of 
the practical role of operations research in business. Because opera- 
tions research is a relatively new field, a brief history of its special 
contribution is given, followed by some examples of applications. 
Its role in the study of business systems is discussed, and the place it 
occupies in the search for a general systems theory is cited. 

BRIEF HISTORY 

Early in World War II, new devices were introduced which 
created vast tactical and strategic problems for England. Radar was 
among these. It had a marked effect on bombing strategy. Sonic 
devices, developed around this same time, made it necessary to re- 
appraise submarine warfare techniques and a host of associated 
defensive and offensive tactics. The global nature of the war created 
complex logistics (supply) problems, the equal of which had never 
been undertaken. 

England formed teams of men whose experience was in the broad 
field of the physical sciences and mathematics. These men were told 
to attack and solve these problems of broad scope. In their work, 
they were to use the tools of the disciplines in which they were 
trained, and were to bring them to bear on the new problems with 
which they were confronted. It was natural that the scientific 
method would be the operating framework for their work. The in- 
sistence on objectivity, measurement of the observable phenomena, 
and the tests of logic seemed to provide the only hope of coping 
with problems for which no tools existed. History has proved the 
wisdom of this step, and an extensive record is available to show how 

156 



Operations research in business 151 

operations analysis, or operations research, became a force in mili- 
tary affairs. 

The United States did notable work in this field during World 
War II, and it is quite natural that it would carry over into civilian 
life in the normal course of events. Thus, since 1946, there has been 
an increasing quantity of material available on this subject. Many 
of the technically oriented professions, especially in recent years, 
have found ways to make use of operations research. Some industries 
have not as yet seen the ways in which it can be put to use. How- 
ever, every month provides new literature on the applications of 
operations research, much of which can be located through the 
references at the end of this book. A large part of the operations 
research literature deals with the application of mathematical and 
statistical techniques to a wide range of problems. In much of this 
work, the scientific method is obvious. When these techniques are 
combined in the solution of real problems, the results may be many 
times more reliable or precise than other less disciplined methods. 

Dr. Philip Morse has expressed 1 the function of operations re- 
search and its methodology: 

In the physical sciences, one starts studying a phenomenon by observing some 
part of its manifold behavior. Next one tries to form a quantitative hypoth- 
esis, a mathematical model of the aspect observed, that will duplicate some 
of its behavior. If one has been clever, or lucky, in his choice of model, its 
mathematical framework will go beyond the observations, will predict what 
will happen in other circumstances. Next comes the phase of controlled 
experiment. One compares the predictions of the mathematical model with 
what actually occurs; if the choice of model has been good, the experiments 
make possible the improvement and strengthening of the model. After com 
tinued alternation of model improvement and further experiment, the mathe- 
matical model becomes a theory, which means the beginning of understanding 
of the phenomenon. This implies an ability to control the phenomenon. If 
the phenomenon as a whole is too complex to work out quantitative relations 
at the beginning, one isolates pieces of the phenomenon, develops a number 
of detailed models, then proceeds to build more and more general theories 
that include the pieces as special cases. 

Operations research makes its most important contribution in 
t he application of mathematics and statistics to business,. Mathe- 
m atical models used by operations researchers are facsimiles or 
representations of real-life p roblems expressed in the form of equa- 
tions. I n effect, operations research employs mathematics to m ake 

1 "Statistics and Operations Research," Operations Research, Vol. 4, page 3 (1956). 



158 Operations research in business 

an a hMiri 1 I i "n i A ^ i i ^l-life p roblem, utilizing a series of mathe - 
matical expressions as the vehicle for stating and solving the prob - 
lem. The value of the mathematical model is that it provides an 
easily changed re presentation of a real-life situation . By inserting 
real val ues into a m athematical model and SQbdng a^d eq uation, it 
is possible to test the operational char acteristics of a system. As a 
testing device, the mathematical model can express the most com- 
plex processes in an abbreviated form. The danger is, of course , that 
the model may easily become to o ^bsfrart, unrrnl nnH b^rr^ ir> ^ 
correct. To be useful, the mathematical model must represent the 
real world. Through the techniques of operations research, it is 
possible to examine the validity of basic premises under which the 
system may be organized prior to any physical commitment of labor, 
material, or capital. The areas of operations research can be sum- 
marized under the headings of probability theory, symbolic logic, 
decision theory, queuing theory, linear and dynamic programming, 
game theory, information theory, and Monte Carlo techniques. 

TECHNIQUES OF OPERATIONS RESEARCH 

Some of the mathematical-statistical tools of the operations re- 
search specialist, are described here. This list is far from complete, 
but it does indicate the wide scope of specific problem solving tools 
which are available. 

Descriptive statistics 

This is a technique of summarization and may be in chart or 
graphic form. Such summaries express relationships between fac- 
tors, and sometimes will show mean values and measures of varia- 
bility such as standard deviation or percentiles. These measures 
present data in different visual modes, some of which may be more 
revealing than others. Through analysis of data presented in this 
way, problem identification begins, and the researcher is able to 
plan his next step. 

Statistical sampling and inference 

This technique guides the researcher in data collection. There 
is always a cost associated with data collection; sometimes it is very 



Operations research in business 159 

high. This technique prescribes the sample size required to be able 
to infer what the total population is like. Obviously, the larger the 
sample, the better; to reduce the error of an estimate about 30 per 
cent requires doubling the sample size. Eventually there is a trade- 
off between cost of data collection and required accuracy for the in- 
tended purpose. Statistical inference can tell whether or not an 
apparent relationship is truly significant, or the result of chance. 

Correlation and regression analysis 

Where statistical inference tells whether or not a relationship 
exists, correlation and regression analysis tell something of the ex- 
tent of the relationship. A relationship or correlation may exist 
between two important factors; but, if the correlation is low, there 
may be no useful purpose that this relationship can serve. Where 
correlation gives the degree of relationship, regression is a statement 
of the equation that exists between the variables. Regression analy- 
sis is a form of curve-fitting with the added benefit that it tells some- 
thing about how good the fit is. Another advantage over ordinary 
curve-fitting is that a many-variabled curve can be fitted in one 
computation, by performing a multiple regression. 

Multiple regression would be useful in the study of the factors 
that determine which of three factories should produce a new prod- 
uct. There are a number of factors that may affect this: time for the 
shipment of product to market, volume of the individual factory, 
level of employment, cost of shipments, or any other of a number of 
measurable parameters. If a number of these variables are chosen 
as the most likely to influence the distance between factory and 
distributor or consumer, then a multiple regression analysis can be 
made of the sampled data. The result will be an equation which 
expresses the relationship between the factors. An examination of 
the equation will show that some of the variables have a great deal 
of influence on the decision of which factory should service each 
distributor, while others contribute relatively little. Thus, the im- 
portant factors and the degree of their importance can be deter- 
mined. 

Multiple regression was seldom used as a business tool until re- 
cently. This is partly due to its complexity and partly to the time 



160 Operations research in business 

required if the problem is a large one with many variables. Com- 
puters have been very useful in solving problems of this type. Large 
scale equipment can handle a multiple regression of almost any 
dimension. Standard computer programs are available from equip- 
ment manufacturers to implement this technique. The availability 
of computer time, proper skills, and appropriate data would make 
it possible to add this type of analysis to the techniques of any or- 



Linear programming 

There are many new applications of this technique, and many 
unexploited but obvious possibilities exist. Optimum locations for 
service facilities can be calculated by introducing this technique of 
mathematics. Analysis of optimum routes for deliveries or collec- 
tions and the associated problems of the location of warehouse and 
factory sites are amenable to solution with this technique. Analysis 
of optimum shipping routes would also be susceptible to this type 
of analysis. 

The large amount of data, number of computational processes, 
and sheer bulk of the problem make it necessary to think of a com- 
puter when there is a linear programming analysis to do. The 
application of linear programming results in more comprehensive 
understanding of the functional relationships of a complex system. 

Factor analysis 

The aim of this technique is to find underlying factors which 
affect the criteria but which may not be well defined before the 
study. An illustration of this would be a study of the effectiveness of 
different departments of an organization and the factors which affect 
them. Here, as in multiple regression, some decision would be 
made as to those variables which are considered most important. 
Sampled information concerning each of these variables would be 
obtained and the factor analysis performed. Initial variables chosen 
for the assessment problem might be organization assignments, 
average age of employee, years of experience (academic and profes- 
sional), measures of proficiency, or any one of a number. The re- 
sulting factor analysis might lead to concepts such as a proficiency 



Operations research in business 161 

factor. This technique is useful in making broad studies. As with 
some of the previous techniques, factor analysis usually requires a 
computer to perform the calculations. 

Control system analysis 

This technique is very powerful when applied to physical systems. 
Its usefulness in business planning would be as a conceptual device 
to aid in developing an adequate model of the system. 

The basic characteristic of a control system is the feedback loop. 
This concept has demonstrated its usefulness in analyzing the opera- 
tions of an "automatic" warehouse. The growth and decay of a 
business may also be expressed conceptually as depending on feed- 
backs from overloaded communication channels, higher cost of 
services, and so on; however, if the specific relationships are not 
known, there can be no control system analysis in the classic sense. 
The availability of a computer to handle the manipulation of data 
would be an asset with this technique also. 

Simulation 

This is the technique which appears to have great promise for 
research. 2 Uses of simulation have grown rapidly in recent years 
largely because of the availability of electronic computers. Early 
simulation studies attacked problems such as inventory systems, 
production scheduling systems, waiting line problems, and so on. 
Recently, this technique has been used for the simulation of an en- 
tire logistics system for the Strategic Air Command. It appears 
entirely possible that a business can also be simulated in a manner 
which will lead to a new understanding of the functional relation- 
ships involved. The important characteristics of the simulation 
technique are as follows: 

(1) Any number of variables can be handled. The only limitation 
is the computer's ability to handle the data. 

(2) The data to be processed can be empirically derived and does 
not have to be smoothed or changed into equation form. 

2 See A. B. Fleet and Company, Chapter 17. 



162 Operations research in business 

(3) The relationship between variables can be complex, i.e., linear 
restrictions do not have to be maintained. 

(4) There is wide flexibility in the choice of procedure; empirical 
data can be used as input and the existing relationships would 
be output. However, proposed relationships can also be used as 
input and the forcasted data will be the output. 

(5) The essential nature of simulation is that the model should vary 
in time, so that the process is a step-by-step reenactment of the 
physical or qualitative system. 

In addition to these technical characteristics, there are some addi- 
tional general advantages which should be noted. Because of the 
large amount of detail which can be built into a problem, simula- 
tion gives results which are useful for the particular system under 
study. Simulation is, thus, less valuable in discovering general laws, 
and would, therefore, be used to gain insight to specific systems. 
Simulation is easily understood and relatively easy to implement. 
The background of individuals performing this activity must be 
broad, but no graduate level mathematics are required. 

Simulation is not practical without a computer, and is completely 
dependent upon the ultra high speed of electronic data processing. 
Thus, a significant characteristic of simulation lies in the computer's 
ability to digest many years of dynamic operation which are ex- 
pressed by the model and compress them into a few minutes of ac- 
tual time. This makes possible the study of a variety of alternative 
paths into the future. The selection of the best path or policy may 
become obvious. But more likely the policy to be followed will 
require a great deal of judgment, and the simulation results will 
aid in balancing one alternative against another. 

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH 

Are systems analysis and operations research the same thing, or 
are they different? Systems analysis at the very least is the method of 
operation used by the operations analysts. As such, it is a part of 
almost every conceivable operations research project. Interestingly 
enough, both industrial engineers and business administration per- 
sonnel have adopted many of the statistical techniques of operations 
research and as a result a larger number of problems have been ex- 



Operations research in business 163 

posed to solution via the "scientific method." There are also some 
good reasons why the systems analyst should be business oriented. 
The business oriented analyst in many ways has better communica- 
tion with his own management. The great expectations of opera- 
tions research have to some extent fallen into the gulf that exists 
between the scientific and business world. The complexity of prob- 
lem solving tools, the vocabulary of mathematics and technology, 
and the inability to translate these into simple ideas has contributed 
to lower realization of some operations research programs. Thus, the 
business trained systems analyst may be called upon to bring the 
tools of operations research to management's attention, to bridge 
the gap of communication. In the process of applying the tools of 
mathematics and statistics to business problems, operations research 
has been limited. The most important limitation results from man- 
agement policy. Each company has the problem of assessing the 
skills it wishes to employ and of finding a group in which the skills 
can be put to work. For the most part, this has meant putting opera- 
tions analysts into staff positions with a loosely defined role. The 
vast majority of businesses are not only unaware of operations re- 
search, but are too small in size to require the full-time services of 
such a specialist. Some few companies have put emphasis on the use 
of operations research tools, the bulk of these in technically com- 
plex areas. 

The scientists who pooled the tools of their various disciplines to 
attack complex war problems provide the model of the mixed team. 
Today it is typical that one man trained in the application of opera- 
tions research tools will be used by many groups who have less than 
a full-time requirement for this special skill. However, other skills, 
like his, may only be required on a part-time basis. Table 9-1 is an 
example of a system study showing three skill requirements, none 
of which are full-time throughout the entire study, and only at one 
point, step 9, are all skills working simultaneously and full time on 
the assignment. The steps of a study are not necessarily restricted to 
those shown in Table 9-1 any more than the skill requirements are 
restricted. These represent one way in which a problem can be 
treated; other skills or other steps are a function of the individual 
problem. 



164 Operations research in business 



TABLE 9-1 

Mixed Team Approach to Complementary Work 
Assignments on a Typical System Study 





Bus in ess Analysis 




Operations Research 


Electronic Data Processing 


1. 


Interviewing 










2. 


Flow Analysis 










3. 


Flow Charting 


4. 


Analysis of Problem 






6. 


Analyze System 
Requirements 


7. 
8. 


Hypothesis of 
Conceptual Model 
Test of Model 


0. 


Analysis of System Data 


9. 


Design of New System 


9. 


Design of New System 


9. 


Design of New System 


10. 


Forms Design 






11. 
12. 
13. 

14. 


EDPM System Design 
Equipment Evaluation 
Program and Code 
New System 
Test New System Routines 



The role of the operations analyst as a member of the mixed team 
can be quite varied. Because of his technical background, he can 
work on equipment as well as non-equipment systems. In either 
case, his interest is primarily procedural and directed at improving 
system performance. His ability to measure performance has ex- 
panded the techniques of system design to a degree of precision in 
system performance seldom realized in the past. In this role, his 
work is basically conceptual and aimed at problem solving rather 
than implementation. The operations analyst brings an objectivity 
and desire for broad studies to business. This is mainly the result of 
his training in a scientific area and his orientation to problem solv- 
ing. The industrial engineer, by contrast, has tended in the past to 
deal with problems in his own area of time standards, plant layout, 
and so on, rather than to take on the broad scale problems. 

Notable successes of operations research in business have been in 
the petroleum industry. The use of linear programming mathe- 
matical models has made it possible to determine the optimum mix 
of raw materials in refinery operations. Using linear programming 
in the solution of distribution problems, the optimum selection of 
the route, and means of transport has returned substantial savings. 
The location of manufacturing or warehousing facilities, custom- 



Operations research in business 165 

arily resolved through the use of engineering economics alone (or 
intuition), has benefited from the use of operations research. 

Operations research has been successfully applied in problems 
requiring the use of sampling techniques. Shop scheduling and 
master planning have been attacked using operations research meth- 
ods, with results that have had practical usefulness for business and 
industry. Freight yard operations, routing of freight cars, cargo 
handling, airline scheduling, and a variety of transportation prob- 
lems have benefited from the use of the operations analyst's tools. 
Queuing models have had wide use in a variety of traffic and service 
problems. In addition, a large number of theoretical problems have 
been attacked, many of which promise substantial payoff in the 
future. The broad scale of these problems emphasize the contribu- 
tion of operations research. It has undertaken the solution of com- 
plex problems, usually with many variables, and attempted to 
structure the system by quantifying the system elements. Because of 
its tools, operations research has brought precise measurements to 
complex problems, and has sometimes obtained for management 
the last few per cents of efficiency in a system which was already 
operating very economically. In many of these cases, the problems 
have been very large in scale, involving millions of dollars in ma- 
teriel, equipment, or personnel. 

OPERATIONS RESEARCH AND A GENERAL SYSTEMS THEORY 

Systems analysis in business may or may not be applied through 
the use of mathematical techniques. Systems analysis and opera- 
tions research share a common methodology by defining an ob- 
jective method of problem solving. However, after ten years of 
operations research in business, there has been no concerted effort 
to provide a general systems theory for the problems of industrial 
management. 3 This is not to say that operations researchers have 

3 Again, H. I. Ansoff, op. cit., writes convincingly about the role of operations re- 
search. He states that it has been directed at problems of the utilization of internal 
resources, not as much at the top management levels. He says it has been dealing 
with problems of "suboptimization" — the process of choosing among a relatively 
small number of alternatives by an administrative level other than the highest. 
Ansoff points out that operations research has made contributions where management 
decisions can be made by computational routines, but that no comparable progress 
has been made in the judgmental area. This latter point is a good argument for 
more activity in the general systems theory problem. 



166 Operations research in business 

done less than they should. If anything, industry itself has limited 
their opportunities. The result, however, is this: Whatever science 
operations research has brought to industry has most frequently 
been in the solution of specific problems, the results of which have 
not been assembled to provide the basis for a general theory. 

Industrial management principles do not provide the requisite 
general frame of reference either. In a few companies, the develop- 
ment of corporate policy and objectives has advanced to a high level. 
However, the analysis of operating problems generally suffers from 
the same drawbacks cited in the example of the pre-World War II 
electrical or hydraulic system (see page 5). A large number of de- 
cisions are reached without cognizance of the appropriate frame of 
reference. Operations research has gone a long way toward reveal- 
ing the dependency of decisions on a large number of variables 
which may be inadequately considered. 

A more fundamental shortcoming occurs in Management's in- 
ability to define its short and long term objectives. In the absence of 
objectives which are carefully interpreted for each level of organiza- 
tion, the uniformity and force of decision making must invariably 
dilute the effectiveness of Management. These shortcomings point 
to the need for techniques by which Management could integrate 
the activities of the organization in an economic, effective way. This 
would give the organization as a whole a unity of purpose from top 
to bottom. 

By recognizing these inadequacies, operations research has con- 
tributed to a recognition of the problems which a general systems 
theory would attack. The continuing efforts of operations analysts 
and systems oriented personnel from all fields will someday produce 
the new concepts and provide the much needed frame of reference. 
It is conceivable that these tools will assist in implementing the 
changes which are required to keep American industry competitive 
in world markets. 



PART TWO 
CASE STUDIES 



10 



G. W. Templar and Company 
(demonstration case) 



On September 1, 1958, George Templar, president of G. W. Temp- 
lar and Company, 1 gave his administrative assistant the problem of 
deciding what was needed to improve their company's production 
control activity. Paul Farber, the administrative assistant, was un- 
familiar with the problem area, but had a working knowledge of the 
company. 

Farber interviewed the production control manager and his as- 
sociates. In the course of a week, he saw men from the areas of 
Factory Supervision, Inventory Control, and Factory Engineering. 
Before undertaking his report, which was due on September 20, he 
planned some investigation in the fields of Sales, Engineering, and 
Accounting. 

After his first week of effort, he listed a series of requirements for 
a successful production control system: 

1. Must be usable and operable on a large scale digital computer 

2. Must be expandable for future increases in volume 

3. Must improve service to the customer by shortening manufacture 
ing cycles 

4. Must decrease inventories by timing the arrival of raw materials 
to arrive close to manufacturing release 

5. Must decrease inventories by providing an orderly system of 
releasing parts into the manufacturing sequence on time 

6. Must decrease unit costs by providing valid rules for scheduled 
factory operation 

7. Must decrease unit cost by supplying a means of knowing ma- 
chine utilization and labor requirements 

8. Must provide a system of in-process controls and feedbacks 

i The purpose of this chapter is to provide a pattern of investigation for subsequent 
cases in the text. 

169 



170 G. W. Templar and Company 

QUESTIONS 

1. What is the problem? 

2. What are some of the questions Templar wants Farber to answer? 

3. What are the functions of a production control department? 

4. What are the objectives of a production control system? 

5. How does the systems concept work in this case? 

ANALYSIS OF CHAPTER I0 2 
G. W. TEMPLAR AND COMPANY 

question 1 : What is the problem? 

discussion: Farber has begun postulating a system before he has made 
an adequate investigation of the situation. His statements are vague 
generalizations that could have been taken from a text book, or the 
opinions of the people he initially interviewed. His statements of 
system requirements are purely qualitative, and as such can be easily 
challenged with careful questions. 

Farber had the problem of learning and reporting about an un- 
familiar area. His idea to interview personnel in the departments 
affected by the production control system, was a good one. There are 
ways in which he might have supplemented his interviews with in- 
formation that would have shed more light on the problem area. Some 
of the things he might have done are: 

(1) Prepare an outline of 

a) what general areas he could cover in the three weeks he has 
to write his report, 

b) what time he would allocate to developing material in each 
of these areas, 

c) how much time he will leave for writing the report itself. 

(2) Prepare flow charts to show major operations of Production 
Control. 

(3) Gather forms, reports and procedures to provide additional 
background material. 

(4) Obtain organization charts to identify the positions and func- 
tions of the various members of Production Control and those 
personnel whose work would bring them into routine contact 
with the production control system. 

2 See Chapter 5. 



G. W. Templar and Company 111 

An assignment of this type is susceptible to the systems approach. 
Using the basic technique of (1) investigation, (2) hypothesis, and (3) 
implementation, Farber would have been provided with an orderly 
way of getting into the problem area. His choice of techniques to de- 
velop the problem area material would to some extent have been 
dictated by what was available for immediate study and how quickly 
he found the most fruitful areas. 

His first undertaking would be to understand the big picture of 
Production Control, and how its operations affected the rest of the 
company. His first problem was, therefore, to establish the objectives 
of Production Control and the boundaries of the problem. Once the 
objectives were identified, Farber would have been able to evaluate 
whether or not the objectives were being achieved. Standards and 
measures of effectiveness of various types would have provided insight 
to the performance of Production Control. If, as a result of this analy- 
sis, there were time left to further evaluate his problem, Farber might 
have suggested some alternative means of improving the performance 
of Production Control. 

Farber might have tried to look at his assignment as a system, as in 
Figure 10-1. Further analysis of how Farber intends to deliver the re- 



LACK OF FIRST HAND EXPERIENCE IN 
PRODUCTION CONTROL; DUE IN 19 DAYS 



TEMPLAR DIRECTION; \ 
DECISION ON HOW \ 

TO ORIENT THIS ASSIGN- Y 
MENT — INVESTIGATION/ 
HYPOTHESIS / 



FARBER 
ASSIGNMENT 



REPORT 
TO TEMPLAR 



CLARIFICATION OF TEMPLAR'S NEEDS; 
REORIENTATION OF ASSIGNMENT 

Fig. 10-1. 

quired report would accent the need for the following: 

(1) Clearer knowledge of the use the report will have. 

(2) Assembling a report which reflects a point of view consistent 
with the time available. 

question 2: What are some of the questions Templar wants Farber to 
answer? 



discussion: There is no evidence to indicate how well Farber under- 



172 G. W. Templar and Company 

stands the assignment Templar has given him. The report which 
Farber must deliver in three weeks cannot contain the results of an 
exhaustive search in areas with which Farber is essentially unfamiliar. 
What, then, does Templar want? Here are some suggested questions 
which Templar may have in his mind: 

(1) Is the production control department properly organized? 

a) Is the staff performing properly all the functions of a produc- 
tion control organization? 

b) Is the department properly staffed? Are there some obvious 
staff weaknesses? Are there surplus or marginal employees 
working in any part of the department? 

(2) What are the problem areas in Production Control? 

a) Are the relationships between Production Control and the 
departments with whom it must operate satisfactory, or are 
there some areas of improvement? 

(3) How does the Templar production control system compare to 
some other system, particularly that of some efficient competi- 
tor? 

This is by no means an exclusive enumeration of all the possible ques- 
tions. Many more will come to mind if this problem is pursued. It is 
important to note, however, that as given, the assignment is quite 
broad and indefinite. This can create many problems for the investi- 
gator. If Farber had pursued the nature of the objectives with Temp- 
lar, he might have made more progress. Farber might have explored, 
for instance, the use to which his report would be put. This would 
have forewarned Farber about Management's objectives. Templar is 
the "systems purchaser" and his needs and requirements should be 
well understood. 

question 3: What are the functions of a production control department? 

discussion: The systems approach is typified by stepping back to take a 
look at the entire system as a whole. This might have forced Farber. 
who was unfamiliar with the area, to ask, "What are the functions of 
a production control department?" Here are some generally accepta- 
ble functions which would have assisted him in his investigation: 

(1) Translate sales department paperwork into practical produc- 
tion plans. 

(2) Keep Management advised on the status of production. 

(3) Maintain adequate records on the factors of production. 

(4) Forecast anticipated labor, material, and machine usage. 

(5) Control the processing of orders in factory production. 

(6) Move materials to the factory floor for processing, between 
operations until completion, and to the shipping department. 



G. W. Templar and Company 173 

These functions explain how the system should operate. But they are 
qualitative statements and suffer the shortcomings of being non-quan- 
titative. However, they are more helpful than the eight system require- 
ments Farber has listed, because they are more general and provide a 
place to start investigating. They do not commit Farber to some idea 
or policy that may be outside Management's frame of reference. The 
analysis of the functions of Production Control might put Farber on 
the trail of a real problem area. 

question 4: What are the objectives of a production control system? 

discussion: Another way of providing an entering wedge would have 
been to ask the question, "What are the objectives of a production 
control system?" Here are some generally acceptable objectives: 

(1) Maintaining schedule — expediting; dispatching; resorting to 
outside production when necessary; predicting overloads and 
underloads; follow-up on outside contracted operations; day-to- 
day contact with factory supervision; assisting in crucial de- 
cision making about priorities and split lots; interpreting the 
schedule and machine load, and so on. 

(2) Coordinating staff services with line operations 

a) Providing the information bridge between the factory and 
the indirect departments such as Sales, Service, Purchasing, 
Material, and so on. 

b) Providing a part of the quantitative feedback information to 
guide staff in decision making, such as the following: 

1) What are the shortages and what are we doing to obtain 
parts or materials? 

2) What short shipments have been received today; how will 
this affect our manufacturing schedule; and when can we 
get additional parts? 

3) What parts are behind schedule, and what are the reasons? 

4) What lots have suffered unusual attrition and are below 
minimum quantities; shall we continue processing at 
higher costs, or should we release an additional lot into 
manufacture and hold the parts in process until the new 
lot catches up? 

(3) Developing and maintaining workable production plans 

a) Providing a system capable of comparing the current and 
anticipated work load to the capabilities of the factory, un- 
der a variety of conditions. 

b) Providing a set of decision rules to be used under conditions 
which are well understood. 

c) Providing the manpower to generate and maintain informa- 
tion essential to reporting the status of manufacturing, and 
planning future operations. 



174 G. W. Templar and Company 

Such points as 1, 2, and 3 above develop the type of data on which 
Farber could base criteria or measures of effectiveness for some part, or 
all of the system. To be effective as criteria, these points must be 
measurable. 

Figure 10-2 illustrates one way to look at Production Control: 



WHAT ARE THE CONTROLS ? 



1. ENGINEERING DATA _ 

2. SALES ORDERS 



PRODUCTION 
CONTROL 



PRODUCTION 
PLANS 



WHAT ARE THE FEEDBACKS ? 



Fig. 10-2. 



(1) Engineering 

a) Controls 

1) blueprint checking procedure 

2) customer approval of working drawings 

3) blueprint release procedure 

4) change incorporation procedure 

5) product standards and specifications procedure 

b) Feedback 

1) change notification and request procedure 

2) engineering order request procedure 

3) master planning notification and change procedure 

4) product standards change request 

(2) Sales 

a) Controls 

1) monthly summaries of customer schedules 

2) copies of master plans 

3) notification of receipt (or absence) of engineering data 

4) sales order data transmission procedure 

b) Feedback 

1) confirmation of scheduled delivery dates 

2) confirmation of changes, cancellations, deletions, addi- 
tions, or alterations in specification to existing orders 

3) reports of status of production 

4) reports of actual shipments 



G. W. Templar and Company 115 

In fact, there are more departments whose activities are related to 
production control operation. The technical requirements of a cus- 
tomer's order involve the tooling, factory engineering, and material 
departments (see Figure 10-3). In each of these subsystem areas, con- 



TECHNICAL 
REQUIREMENTS 







TOOLING 






DATA 
















ENGINEERING 
DATA 





TOOL 
PLANNING 



MATERIAL 
PURCHASE 



PURCHASE 
•ORDERS TO 

VENDORS 



Fis. 10-3. 



trols and feedbacks are required to provide for accurate and reliable 
transmission of information. Assign individual areas for more com- 
plete diagramming. 

question 5: How does the systems concept work in this case? 

discussion: Systems postulation starts in the investigation phase. It is 
impossible to avoid thinking of a problem area without also consider- 
ing what to do about the problem. However, a requisite to systems 
postulation is the availability of pertinent data, particularly of an 
operational and quantitative character. Here are some of the typical 
data which would be required in each of the categories Farber has set 
up before systems postulation could begin: 

(1) Must be usable and operable on a large-scale digital computer. 

a) What data is being processed? 

1) How frequently? 

2) What data must be transmitted? 

3) What subsystems are affected? 

4) What basic files and transaction files are concerned? 

5) What is the compatibility of the subsystems being proc- 
essed? 

6) How many items per file are being processed? 

b) Who is processing the data now? 

1) Who originates? 

2) How many copies? 

3) Who gets them? 

4) What do they do with them? 

5) What is their ultimate destination? 

6) What equipment is currently used to accomplish the data 
processing in question? 



176 G. W. Templar and Company 

7) Are the forms and procedures for this data processing 
application operating as they were originally designed? 

8) Are the personnel originally designated for this data 
processing application actually doing the work? 

c) What are the costs associated with this data processing appli- 
cation? 

1) Is cost information available? Incomplete? Current? 

2) How are costs allocated or assigned to this procedure? 

3) Are the costs as they were assessed when the procedure 
was designed? 

(2) Must be expandable for future increases in volume. 

a) What were the original volume figures? 

1) What are the current figures? 

2) What are the projected figures? 

b) How do the costs per unit transaction bear out the need for a 
flexible system? 

1) Will an expansion in volume require changeover to more 
complex or costly machinery? 

2) What will the effect of a shrinkage in volume be on the 
cost picture? 

c) Can revisions in the existing system eliminate the problem 
of expansion or shrinkage in the volume of transactions? 

(3) Must improve service to the customer by shortening manufac- 
turing cycles. 

a) What means are at the disposal of the company to shorten 
the cycle? 

1) Decrease lot size? 

2) Increase automation? 

3) Re-engineer or re-tool for easier or more economic manu- 
facturing? 

4) Reduce quality? 

5) Loosen specifications? 

b) What are the side effects that may accompany the move to 
shorten manufacturing cycles? 

(4) Must decrease inventories by timing the arrival of new materials 
to arrive close to manufacturing release. 

a) Who originates the order to buy materials? 

1) Who processes the requisition? The purchase order? 

2) How do they know when to requisition? 

3) How do they know how much to order? 

b) Who gets the purchase order? 

1) Is a choice among several vendors made? 

2) How is the purchase order prepared? 

3) Who prepares it? 



G. W. Templar and Company 177 

4) Is the time required to prepare a purchase order incorpo- 
rated in the lead time for manufacturing? 

5) How is contact maintained with the vendor? 

6) How is the due date of a delivery determined? 

7) How does the system respond to a late, early, short, or de- 
fective delivery? 

c) What is the current level of inventory compared to the vol- 
ume it will support? 

1) Does quantitative data indicate inventory is too high (or 
too low) for anticipated deliveries? 

2) Is there competition or trade information bearing on the 
problem of inventory level? Turnover rate? 

3) Does management agree inventory level is satisfactory? 

4) What kinds of material costs are kept? 

a. What is the accuracy and reliability of records? 

b. How does a sampling of stock items check with 

1. quantity shown in inventory records? 

2. description shown in physical inventory records? 

3. location files? 

4. cost files maintained in accounting and physical 
inventory records? 

(5) Must decrease inventories by providing an orderly system of 
releasing parts into the manufacturing sequence on time. 

a) Will this decrease inventories? Increase inventories? 

b) What measures can be devised to support the need to oper- 
ate with less (or more) inventory? 

c) What symptoms exist to reflect an improper level of inven- 
tory? 

1) Orders going into process late? How many? Did they catch 
up? 

2) How many orders are required in the pipeline? At what 
stages? 

3) How many releases are now in stock for some sampling 
of our current production schedule? 

4) How many geographically separate stocks of identical 
parts exist? 

5) How many lots are 

a. in process? 

b. at outside contractors being processed? 

c. in transit? 

d. on order? 

6) How do the lots in process and on order compare with 
the schedule requirements for some sample group of 
parts? 



17 S G. W. Templar and Company 

7) Is there a shortage list? 

a. How many items? Have they continuously recurred? 

b. Are the reasons for shortages known? Are they being 
followed by Production Control? 

c. Who issues the shortage report? How often? Who uses 
it? What is its reliability? 

(6) Must decrease unit costs by providing valid rules for scheduled 
factory operation. 

a) is there an existing system with decision rules, control, and 
feedback — or is this a collection of non-integrated proced- 
ures and processes? 

1) Who designed it? Who polices it? 

2) Is it up to date, or half used and half improvised? 

3) Are the forms standard or bootlegged? 

b) Is the system followed? 

1) What are the deviations? 

2) Is there provision for changes? 

3) Has it been changed since inception? 

c) Will this decrease unit costs? Increase? 

1) What is the relationship of scheduling rules to cost? 

(7) Must decrease unit costs by supplying a means of knowing ma- 
chine utilization and labor requirements. 

a) What controls exist on the factory level? 

1) Reliable timekeeping? Who does it? How many time- 
keepers? What do the timekeepers do? 

2) Are department and cost center lines clear? 

3) Who moves parts? How many dispatchers? 

4) Who counts parts? Who records quantities? "Where are 
quantities recorded? Is parts count checked? 

5) Are there rules for lot splitting? Are they applied? 

a. Who has the authority to split lots? 

b. Is lot splitting the exception? What data is available? 

6) How are employees transferred between departments? 

a. How are labor changes handled? 

b. How does a foreman know when to borrow and when 
to release? 

c. How are absent employees replaced? 

b) How is shop loading accomplished? Is it a "system"? 

1) What is the frequency of job changes? 

2) Are methods standardized? Are the work standards repre- 
sentative of the work being accomplished? 

3) How far in advance is the work load known? 

4) What is the frequency of reporting completed work? 
a. In what detail is completed work reported? 



G. W. Templar and Company 179 

5) Are there machine classes? Machine numbers? 

a. How is a reference in the schedule tied to a particular 
machine in a department? 

b. How is a particular lot in the schedule tied to a lot in 
the factory being processed? 

(8) Must provide a system of in-process controls and feedbacks, 
a) Where is the current system failing? 

1) Absence of accurate inputs? 

a. Standards? 

b. Factory routings? 

c. Machine assignments? 

2) Are processes too slow? 

a. Files not updated? 

b. Reports delivered too late to be useful? 

c. Performance information incomplete? 

d. High cost of making essential data available? 

3) What are the controls in the various systems? 

4) What are the feedback mechanisms? 

Much more could be written under each of the above eight head- 
ings. It must now be obvious that an investigation of the size necessary 
to support the system requirements Farber has stated, is far outside 
the realm of the time available and the assignment as Templar had it 
in mind. It is also obvious, that qualitative, non-operational state- 
ments will not stand up unless supported by quantitative data. These 
questions begin to develop the line of investigation which will reveal 
the type of quantitative data which is desirable, if a full blown in- 
vestigation is required. 

SUPPLEMENTAL DISCUSSION QUESTIONS 

1. Why was Farber picked to do this job? Who else might have been 
assigned this problem? 

2. What would be suitable criteria of a production control system? 

3. Has Farber's job been well defined? Who has the responsibility 
for stating the problem? 

4. What should Farber's report contain? 

5. At what point should the postulation of a system begin? 



11 



The Lee Company 



The Lee Company has been in business in Leominster, Massachu- 
setts, for 42 years. During this time they have built a reputation as 
one of the foremost manufacturers of machine shop parts in the 
country. Plant locations are in New England but two plants are in 
the deep South. Direct labor at the Leominster plant numbers 
1,200. Total direct labor employees of the corporation number 
3,700 at all plant locations. 

The Lee Company has several competitors of equal size through- 
out the East. Machine shop know-how is the principal criteria of 
a competitor's ability to undertake manufacture. The variety of 
products is narrow, but the number of models produced in each 
category creates a long and impressive list of end products. The 
total number of models active at Leominster at any one time might 
average 300. Each model consists roughly of 25 manufactured parts 
and 80 purchased parts. Each manufactured part has an average of 
25 operations. Customer orders at any one time might call out as 
many as 60,000 end products. 

The facilities of the Leominster plant contain offices, shipping 
and receiving, stockrooms for raw materials and finished goods, 
machine shop areas, and assembly areas. The scope of activities at 
Leominster is sufficiently wide to necessitate a full organization, 
both direct and indirect. The direct departments are seven in 
number, each headed by a Foreman who has an assistant on each 
shift. The six machine shop departments vary in size between 50 
and 200 people. The manufacturing areas in each department are 
divided into sub-areas called cost centers. Costs are accumulated 
by employee, by cost center, by department, and by division. 

180 



The Lee Company 181 

The Lee Company had its greatest growth from 1943 to 1955. 
While in no sense a "war baby," Management seized upon the op- 
portunities offered by government financed business to expand 
their resources threefold over this period. The effects of this expan- 
sion not only created manifold plant problems, but personnel prob- 
lems as well. The production control department was organized 
along the lines of what the government demanded of industry dur- 
ing war time. Because firm commitments were required of any 
major subcontractor, Management decreed that "something else" 
was needed between Sales and Production to coordinate the expect- 
ancies of one against the day-to-day problems of the other. Figure 
11-1 is a schematic drawing of the Lee Company Production Sys- 
tem. 

Production Control personnel were recruited from Sales and 
Production. In addition, specialists from other companies were 
recruited and several people were brought in, over a period of time, 
from Factory Engineering and Industrial Engineering to comprise 
the staff. The full staff of operating and clerical personnel in Pro- 
duction Control eventually grew to 85 persons working on two 
shifts. 

At the outset, Production Control carried the psychological bur- 
den of white collar personnel working and making decisions in the 
production area. The first efforts of Production Control to gain con- 
trol over production were not effective, and yet some opinion on the 
staff level held they were working in the proper direction. Initial 
undertakings were concerned with determining what reports Man- 
agement would require, what reports would be necessary as a means 
of achieving this end. 

The end product of several years of effort were the Customer 
Schedule (summary of all customer orders) and the Production 
Requirements Report (purchased parts summary). These were the 
key reports in the purchasing and manufacturing process and the 
timeliness of each was considered extremely important. Accuracy 
was a major requirement. These reports were issued monthly by 
the Tabulating Department from punched card decks. For instance, 
new customer orders were key punched on acceptance, and ship- 
ments were key punched on receipt of a shipping notice. These 




Fig. 1 1-1. The Lee Company production system. 



The Lee Company 183 

inputs were merged and matched with the card decks representing 
the Customer Schedule. 

The Customer Schedule was made available about the tenth 
working day of every month. The other schedules were published 
about the twentieth working day of every month. Although there 
was continuous discussion about the lateness of schedules, Produc- 
tion Control argued that the projection of several months on each 
schedule release was adequate to cover any time loss in producing 
paper work. 

The Combined Release Schedule was the result of multiplying 
the Customer Schedule by the Parts List of an end product and sub- 
tracting any unallocated work in process or finished goods. Since 
the manufacturing cycle was four months and the porcurement 
cycle was two or more months, it was deemed advisable to carry the 
requirements out to ten months. This schedule showed numbers 
of parts plus scrap allowances as well as indicating the status of work 
in process. Back Orders represented the extent to which the com- 
modity on order was behind schedule. 

From the Production Requirements Report, Production Control 
requisitioned the purchasing department to buy the net require- 
ments of purchased parts and raw materials. The problems of pro- 
curement and timely delivery of raw materials were then the func- 
tion of Purchasing. 

Simultaneous with the release of raw materials into the produc- 
tion process, the Parts Production Routing was sent to the operating 
departments concerned. These became the processing "bible," and 
carried all operation numbers and information which would be 
required to complete each detail. Areas to be machined were indi- 
cated on each blueprint by heavy lines. Engineering changes were 
supposed to be incorporated in blueprints prior to production. In 
periods when changes were frequent, end products were serialized 
and changes were incorporated at irregular intervals. Changes in 
routing were reported to Factory Engineering, if, in the opinion of 
the Foreman or Dispatcher, the changes were permanent. These 
routings, in addition, contained pertinent information which 
would assist the operating department in manufacturing planning. 

Although efforts to describe the release date of raw materials in 



£2 

co 



*3 



O 
OS o 



I 

u 



« rH 

§ to 

o o 

H rH 
CO 



W 
ft 



, -o 



Oh O 

W to 
H 



O" o 



CM CN CM >T\0 



r, * W 

H O Q 
W < O 



g H 

<8 



ONOOO 
v£> CMvO (-IvO 



^ rH ir\rH^O 
-*r-H O f\ 






tr. 



ft 

H 
W 

O - 



•00 tO tO CNrH 
H i-HrH l^C- 
rH tO tO HO 
OOHWN 



ooooo 
ooooo 

CM CM CM CM CM 



O N rHsOO 

vO <t>rH r-\ CM 

SO _* O 

C*- to ir\ 



vOOO 

HOO 



0>\ 






OOOfcJE-" 

Enh^OW 

«c •<«; wm 

H MMO<C 
CO COCO 

I I 

.J 
•J 

w 

X 
CO 



HHHfAt- 



OS 



OS 
«< 

oco 



C-\C—C--C-C— 



O Ol 00 O (M 

i>- co Q -h — i 

r-H O © CO 

lO CM 



xT> O 00 O 00 — 

O O CO — i lr- \T) 

cm cm o co cm 

-i 00 — < 



6 T= 
-w, to 



O 05 00 
!>• CO O 

m CM 



CO 


CO 


o 


o 


00 


o 


to 




o 


o 


CO 








CM 


CM 


00 


CO 






^h o *n 
o o o 

O !>• CO 



8 



o 

CM 


t> 


o 


§ 


-* 




!>• 


co 


"-* 




O 








O O 00 -h 

00 -"f O CM 

— i O 1> CO 

CO CM 



co 


00 


liO 


_ 


oo 


o 


oo 


_* 


CO 


o 


CM 


r— 


t^ 




£- 


co 




00 


CO 


CM 


CO 


3 


CM 








^ 


1—1 


00 




"—i 





»OiO^O»OOiOiOvOinoOOOCOI>-CMO 



CO 1^ CO m CO 00 CM 



CO CO t-. 



z 
< 


M 
P 

Q 

UJ 






*e 


-f 


O 


iO 


^ 


»o 


O 


Oh 




to 


to 


iO 


00 


00 


CO 


TH 


tH 


s 


— 




e 


to 




CM 


CM 




l-H 


—i 




u 


u 
tw 




Q 


to 


CO 


00 


00 


CM 


GO 


cm 


h-1 


as 

























i* 




to 


o 


! 


CM 


oo 


00 


cc 


H 


to 


to 

■g 


00 


o 

CO 


CO 


© 


"*l 


CM 








t— 


<* 


■^ 


!>■ 


J>» 


rJH 






3 


3 

s 


CO 
CM 


< 


< 




o 

CM 








o 



















CO tH rH 
Oi O — i 



Q °° 
co p4 



00 £-. 



m to 



00 00 Oi 00 CM 



—i i-H 00 fH ^H 00 Q 

O «-h O 00 — i O H 

—< CO i-H ^h CO 00 °? 

co o — i oo ^ m £r 

cm co oo r-» co co ^ 

W3 



to s 

ts-< 



O oo 

t-- 00 

00 *H 

CM CO 



* I 

s 



-f- 


_! 


v^ 


__ 


CO 


_ 


i-. 


o 


l> 


o 


X 


o 




o 


TdH 


00 


as 


cc 


00 


00 


cc 




CO 


~-* 


P-H 


■<*■ 


~r 


~ 



cc 


00 


rH 


>o 


>". 


cc 


00 




o 


© 


1^ 


o 




iTi 


cc 


to 


00 


CM 


t^ 




CO 


© 


© 


00 


O 



to 

53 
S 
&. 

to 
g 
O 
co 


cj 

i 

V 

— 

"c3 




U 

% 


-J 


o 

U 

1 

o 


CJ 

C 

c 
cu 

CJ 

o 

C 

> 


Oh 
O 

u 


o 


o 

CU 
CJ 

U 

cc 


6 
u 

ft 

"> 
JZj 


Oh 

t-H 



u 

g 

t?5 

c/i 




U 

Ja< 
O 

■d 
c 


S-H 

O 

u 

— 

CJ 

< 

h 


6 
u 

"3 

CU 

O 



o 

U 

CJ 

op 


C 
c« 


o' 

U 

u 

V 

cr 




u 


c« 


<y? 


tu 

C/5 


s 


IS 


< 


C<3 


t3 


-C 


CQ 


O 


H 


J 


h3 



s 



Cf VC C*l 
— X — 



t> so t- O 

OC — SO so 

~. x ~ — 






l> x"- 

CM 03 



— X O CM SO TT t^ 

gj so x cm — 

so c\r — 



— ~ x t> — 

so X to ~r ^ 
CM so X (M 



.*"* -2 -^ -5* '"' 



i-o — — x — J^ cm 

^r tt x c^ cm cm 

:>. io ^ -o so 

IM •* O — — 









cm 



IN 



CO _ L-W^ IN - C S L * X 

ta Z* m cm " — ^f 

— O IN »n 



x — — x t^ — ^r i-o 
— cmxc: — t^TX 
t-ooo<o*^ooo 

t**cO— ■ lOOt^OOO 



x o to 



H 2 



Tt< tjh Th 

CM CM CM 
t^ f-~ CO 



O O O 

o o o 



00 00 00 00 



© o o 

© © © 



© © © © 

CM CM -* 1 O 
CO GO CM "*• 



00 00 CO Tfl 



© 

CO 
CM CM CM 



-h CM 



CM i-h 


OS 


00 © 


Ol 


o> 


CM TjH 


00 


00 00 


00 


00 


t^ © 


1— 1 


i—i 


l"H 


FH 



1— 1 


© 


CM 


© 


© 


i— 1 



§ 



-I 

^5 



$ 






CO wM 



8! 



— I GO Q © t^ 

00 7-1 © © 00 

I>. CM rf< -h © 

•M r-H CM 



hJ -k, "T2 



© © © 

CO "f TJH 



Tf CM ^O CM 

lO © -H © 
Tf © ~> -H 

iri Tti CO >— i 



CO lOiO 

t^ r^ r~- 

-<**©© 



co *o — i rt< 
t^ © CM CO 
Tt< CO -*i © 



"* CO H 



© CO © ^ © 
t^t>.t^CM r^ — « i—< i— < 



00 


oo 


rh 


§ 



188 The Lee Company 

terms of the due date in stock of finished parts (called lead-time or 
set-back) had only begun, Production Control was insistent that 
more accurate set-backs would solve many of the problems of irreg- 
ular production and long manufacturing cycles. Ganntt charts illus- 
trating the relationship of "in dates" compared to "due dates in 
stock" were made for a few end products. It was noted that setup 
time plus machine time was about 25 per cent of the elapsed 
calendar time the part was in process. Man hours for machine setup 
and processing were produced on Machine Load Forms (Figure 
11-3). Standard hours were adjusted to reflect actual anticipated 
hours that would be consumed. 

Start and due dates were taken from the Manufacturing-Calender 
Day Cross Reference Chart. The use of this chart made it possible 
through simple addition and subtraction, to determine over a three 
year period, the in and out dates of any part or operation in the 
manufacturing schedule. There were significant problems in keep- 
ing this system up to date. 

The production control department and the industrial engineers 
assigned to the machine shop departments met to review a tentative 
proposal (see Exhibit One) that would form the basis for a common 
understanding on scheduling in August, 1956. After some discus- 
sion, the report was set aside pending a reopening of the entire 
scheduling problem. The opinion expressed in this meeting con- 
firmed the mutual agreement on the need for more system. 

Mr. Albert King, Chief of Production Control, made this state- 
ment in a meeting held September of 1956 in the Division Man- 
ager's office at Leominster: 

Production Control at Leo has turned from a control function to one of fire- 
fighting. I doubt if we have ever had control, or ever will have. The tab 
records and our work in process records don't tally. Our dispatchers are con- 
stantly bailing out some promise of our Sales Department; our counter-movers 
don't count and our rules of thumb for release of detail parts into the system 
are inadequate. I question the decision rules employed by the factor)' super- 
vision, and Industrial Engineering gives us no practical data to support our 
plan for machine loading and labor utilization. We have no concept of lots, 
optimum lot size, or identification of lots in process. The documents that 
keep our system going are the scraps of paper pieced together by the Dis- 
patcher which tell him which jobs are hottest and how high the temperature 
of each hot job is, compared to the others. Our shop is no longer a production 
shop, but a custom shop, and has been for a long time. 

























CO \ 


a 
< 




^ 
^ 


^~§ 


^ 


^^ 


5> 

\3 






w 


<^ 


S3 


fc* 


*> 


< S 


\ 


^ ^ 


N 






2 
z 


1 

3§ 


3 












* 




|li 

< z w 






^^ 


^ 




^s 

^ 








"<K 


tS 2 ft 

co jjjj 












^ 




6 
z 
h 

ft 

a 

o 


ft 

1 




w 2 

co «^ 




^5 


^^ 


^^^ 


55 




d 
w 
a, 
w 
,-1 
3 


d 
z 

H 

Ph 

w 

Q 






^1 


1 




5? 






09 

U 


W 

H 










1 










">s^ 


ft en 


5 § 


Ss 


^ $> 


^ 


^ ^ 


^ 






z 


V 


& g 


^ 


v. ^ 


\ 


^ °\ 


K 








s 

^ 


CO « 


N 


•* 








^ 






^ ^< 


X 


\s >5 


\i 


^ ^ 












u 

3 


*& 


$ 




s! 








< 


6 




Ctj 


^ <\ 


^ 


<\ ^ 


J\ 


£> 5^ 


^ 






H 


Hj 




<r <: 


( \ 


N °y 


N 


^ V 






si 

ft 

w 


^ 

^ 
^ 


o 
















a q 




X 




V \ 


\ 










a 

o 

H 
CO 




















Sh 


^ ^ 


rv 


^V CV/ 


^N 


\a *s 


^ 


/ 




u 


«a 


\s f\ 


r\ 


<\ ^ 


c 


^ ^ 


^ 


/ 


w 

H 

Q 


fc 




«$ 


K. 


^ a 


^ 








o 
z 




s 














t-l 




\ 




< 

Q 

■4 
w 


S 


w 

Q 

o 
u 


Q 


<<< 












\ 




pcj 




< 

cr; 
w 
H 

< 

a 


Q 
O 
O 
O 


i 














z 
S3 

a 
z 

H 






CO 






















z 






















< 


^ 


& Pj 


^ 












/ 




:.y 


^ 


&< 


^ 












/ 




H 


^ 




\ 














o 
z 

H 


^ 


73 

& k ■* 


















O 


X 


5 ° a 














/ 





















£ 
o 

U 



on 



6D 



190 The Lee Company 



MANUFACTURING-- CALENDAR DAY CROSS REFERENCE CHART 
BOLD FACE NUMBERS = MANUFACTURING DAT 



1956 



JANUARY 


S 


M 


T 


W 


T 


F 


S 


1 


2 


3 


)( 


5 


6 


7 






1 


2 


3 


4 




8 


9 


10 


11 


12 


13 


Hi 




5 


6 


7 


8 


9 




15 


16 


17 


18 


19 


20 


21 




10 


li 


12 


13 


14 




22 


23 


21» 


25 


26 


27 


?8 




15 


16 


17 


18 


19 




29 


^o 


31 












20 


21 










FEBRUARY 








1 


2 


3 


li 








22 


23 


24 




5 


6 


7 


8 


9 


10 


11 




25 


26 


27 


28 


29 




12 


n 


Hi 


15 


16 


1? 


18 




30 


31 


32 


33 


34 




19 


20 


21. 


?? 


23 


2k 


25 




35 


36 


37 


38 


39 




26 


?7 


28 


29 










40 


41 


42 








MARCH 










1 


2 


3 










43- 


44 




U 


5 


6 


7 


8 


9 


10 




45 


46 


47 


48 


49 




11 


12 


13 


11+ 


15 


16 


17 




50 


51 


52 


53 


54 




18 


19 


20 


21 


22 


23 


214 




55 


56 


57 


58 


59 




25 


26 


27 


28 


29 


30 


31 




60 


61 


62 


63 


64 




APRIL 


1 


2 


3 


h 


5 


6 


7 


65 


66 


67 


68 


69 


70 




8 


9 


10 


11 


12 


13 


lit 




71 


72 


73 


74 


75 




15 


16 


17 


18 


19 


20 


21 




76 


77 


78 


79 


80 




22 


23 


2U 


25 


26 


27 


28 




81 


82 


83 


84 


85 




29 


30 














86 













Fig. 1 1-4. 



EXHIBIT ONE 



MAY 


S 


M 


T 


W 


T 


F 


S 






1 


? 


3 


h 


5 






87 


88 


89 


90 




6 


7 


a 


9 


10 


11 


1? 




91 


92 


93 


94 


95 




13 


Hi 


1^ 


16 


17 


Ifl 


19 




96 


97 


98 


99 


100 




20 


21 


?? 


?3 


?h 


?■" 


?6 




101 


102 


103 


104 


105 




27 


?a 


29 


30 


31 








106 


107 




108 






JUNE 












1 


2 












109 




3 


u 


5 


6 


7 


6 


9 




110 


ill 


112 


113 


114 




10 


11 


1? 


13 


Dl 


1^ 


16 




115 


116 


117 


118 


119 




17 


18 


19 


20 


21 


22 


23 




120 


121 


122 


123 


124 




2ii 


25 


26 


27 


23 


29 


30 




125 


126 


127 


128 


129 




JULy 


1 


2 


3 


h 


5 


6 


7 




130 


131 




132 


133 




8 


9 


10 


11 


12 


13 


Ml 




134 


135 


136 


137 


138 




15 


16 


17 


18 


19 


20 


21 




L39 


140 


141 


142 


143 




22 


23 


2b 


25 


26 


27 


28 
















29 


30 


31 












144 


145 










AUGUST 








1 


a 


3 


u 








146 


147 


148 




5 


6 


7 


8 


9 


10 


11 




149 


150 


151 


152 


153 




12 


13 


1U 


15 


16 


17 


18 




154 


155 


156 


157 


158 




19 


20 


21 


22 


23 


?h 


?5 




159 


160 


161 


162 


163 




26 


2? 


28 


29 


30 


31 






164 


165 


166 


167 


168 





SEPTEMBER 


S 


M 


T 


H 


T 


F 


S 














1 
















2 


3 


L 


- 


6 


7 


fi 






169 


170 


171 


172 




9 


10 


11 


V 


13 


T| 


'- 




173 


174 


1 75 


176 


177 




16 


17 


16 


19 


20 


2L 


22 




178 


179 


ISO 


181 


182 




25 


2u 


25 


26 


27 


28 


29 


// 


183 


184 185 156 


187 







1 


? 


• 


li 


«f 


6 




188 


189 


ISO 


191 


192 




7 


8 


9 


10 


n 


1? 


13 




193 


194 


195 


196 


197 




Hi 


K 


16 


M 


18 


19 


20 




198 


19? 


20 m 


201 


202 




21 


22 


23 


2:. 


25 


26 


27 




203 


204 


205 


206 


207 




28 


?9 


30 


V: 










208 


20? 


ZM 









ggggjBig 











1 


? 


3 










211 


212 




)l 


«? 


6 


7 


8 


9 


10 




213 


214 


215 


216 


217 




11 


12 


13 


111 


15 


16 


17 




218 


219 


220 


221 


222 




18 


19 


20 


21 


22 


23 


21+ 




'223 


224 


225 




226 




25 


26 


27 


25 


29 


30 






227 


228 


229 


230 


231 





DECEMBER 















1 
















2 


3 


u 


5 


6 


7 


8 




232 


233 


234 


235 


236 




9 


10 


11 


1? 


13 


Hi 


"K 




237 


238 


239 


240 


241 




"|6 


17 


18 


1- 


?0 


?1 


?? 




242 


243 


244 


245 


246 




2*0 


Mx 


?<; 


?r 


27 


?« 


?9 








247 


248 


249 





NOTE ON SCHEDULING 



I. The ideal flow situation occurring in mass production: 
A. Automation: Material handling and labor minimized. 



The Lee Company 191 

B. Progressive Assembly: Material handling and labor optimized 
on another level. 

1. In a job shop, the ideal is the closest approach to progressive 
assembly as can be obtained without sacrificing flexibility, 
a. One way to do this is to standardize on the actual hours 
per lot. To do this, a mean time would be derived which 
combines setup cycle, production, and lot size in one value. 
Very little variability above the mean time would be al- 
lowed, since this would delay other lots. 

II. The delay that results from unbalanced production: 

A. Unbalanced production results when mutually dependent oper- 
ations are delayed by waiting. 

1. The causes of delay between operations: 

a. Wide distribution of lot sizes 

b. Wide distribution of standard hours per unit of operation 

c. Wide distribution of setup time value 

d. Other causes are excluded for the purpose of this note 

2. Waiting delays are interdependent: 

a. A delay on any one part forces the same delay on all parts 
following — a larger cumulative delay builds up as parts 
get closer to final assembly. 

b. Minimization of waiting delays will reduce total make 
span 

III. Waiting delays — a series of distributions capable of analysis: 

A. Data are available or can be made available to analyze delay. 

B. The sum of waiting elements is a time measure. 

C. The number of lots having priority over the n th lot to be proc- 
essed also have a distribution. 

D. The value of the waiting time measure, times the quantity of 
lots to be processed before the n th lot can be put on the machine, 
describes the waiting time and machining time for the ?i th lot. 

1. Wide variations on the high side must be compressed by 

a. Better tooling (to reduce operating cycle) 

b. Better work methods (same) 

c. Duplicate or triplicate tooling (so two or more machines 
do the same jobs) 

d. Use of different machines 

2. Other variations may require changes such as 

a. New setup procedure 

b. Selection of different lot size 

IV. The use of the waiting measure: 
A. As a unit of production. 

1. Lot sizes are conditioned by the requirement that the total 



192 The Lee Company 

time to do any setup and operation must be a value within 
certain limits. This will provide an immediate return by 
minimizing delay. 

B. When combined with other measures, it has the capability of 
describing a finer setback. This application has future impor- 
tance and is not timely now. 

1. Other elements of the setback, such as transit time and ran- 
dom delays, must be added. 

2. Operations performed outside the department must be ana- 
lyzed in the same way. 

C. Day-to-day situations in actual operation would benefit from 
any plan that would tend to equalize the number of hours am 
machine must run on any operation. 

QUESTIONS 

1. What is the problem? 

2. Where would you begin in this task of making Production Con- 
trol an effective organization? 

3. Evaluate the proposal on scheduling. 

4. What are the major subsystems required for an integrated Pro- 
duction Control? 

5. In each subsystem, find the inputs, outputs, controls, and feed- 
backs. 



12 



Marxson and Company 



Bill Heilson was a member of the Accounting Department of 
Marxson and Company. 1 As a senior member of Management, he 
was regarded as the "trouble-shooter" for a wide variety of prob- 
lems. This activity was much in evidence in the design of punched 
card systems, tabulating routines, and cost accounting procedures. 

Because he had an interest in electronics and was highly regarded 
by the corporate management, he was invited to attend a meeting on 
June 12, 1957. Sales representatives of a well known electronics 
manufacturer were scheduled to make a presentation on electronic 
data processing equipment for business. Heilson was well read on 
this subject and had encouraged members of the accounting depart- 
ment to learn about this new tool. He had no direct experience, 
however, with an operating computer installation. 

The electronic computer manufacturers were well prepared for 
their presentation. They had spent several days working with their 
local sales representatives who serviced Marxson and Company, to 
personalize their talk on computers. Since Marxson was a multi- 
division manufacturer of ladies ready-to-wear, the presentation in- 
cluded examples of how other firms in the same field were planning 
to use electronics in business data processing. There were no com- 
pleted installations to which the computer manufacturers could 
point; however, there were many companies either investigating or 
studying their data processing problems. Because this was an initial 
meeting, costs and savings were the principle themes of the presenta- 
tion. 

Henry Marxson, President of Marxson and Company, was deeply 

1 See Chapter 6, page 107. 

195 



194 Marxson and Company 

impressed by the advantages that high-speed data processing would 
bring to his firm. As a result, he held several corporate level meet- 
ings in the following month. Marxson gained agreement of his 
officers to launch a formal investigation to determine if computers 
would be practical for his company. On July 20th, Marxson asked 
Heilson to form a committee whose purpose would be to investi- 
gate this problem. On August 1st, the committee was formally an- 
nounced with Heilson acting as Chairman. Other members of the 
Committee were the principal members of top management, in- 
cluding A. M. Biddle, the Treasurer. 

The committee met twice monthly to discuss the problems of 
installing electronic data processing. Each meeting brought out 
questions which required Heilson to spend time making interim 
investigations. Some questions had to be shelved because they were 
either dependent on future undetermined factors or were of a 
technical nature. It became obvious at an early stage that expert 
advice on this problem would be advantageous, despite all of the 
training, education, and exposure it might be possible to bring to 
the staff who would be involved in the studies. A possible alterna- 
tive they considered was to hire a top level, technically oriented 
expert to head the study team. 

Heilson invited all of the manufacturers of medium and large- 
scale computers to visit him at his office in Milwaukee, Wisconsin. 
Over a period of two months, a substantial amount of literature 
was collected and surveyed. During this time, the computer manu- 
facturers talked with Heilson. Most of these firms offered, as a part 
of their services, manpower to analyze some of the primary data 
processing systems. However, it appeared that time and cost would 
not permit an equipment manufacturer to make a thorough analy- 
sis of the business data processing systems, some of which were 
manual and others, such as payroll, on punched cards. Furthermore, 
it appeared logical that the computer manufacturer's role would be 
to sell equipment, not business systems. 

All of the manufacturers stressed economy of operation. One of 
the arguments for computers was that electronic equipment over- 
came the necessity to do extensive hiring or layoff of data processing 
personnel, as business volume rose or fell. Companies were advised 
to consider equipment with sufficient capacity to take on additional 



Marxson and Company 195 

data processing burdens and flexibility to operate almost as econom- 
ically, on smaller loads. Because of the complex nature of electronic 
equipment and the way in which it operated on data, the need to 
design business systems with the computer in mind was also a major 
consideration. 

Heilson suggested to the Committee, in October, that it would 
be wise to investigate the use of consultants to assist in the studies 
leading to the selection of equipment. Some members of the com- 
mittee agreed that this might be necessary because each computer 
had distinctive technical and operational characteristics. Heilson 
insisted that the need to compare equipment characteristics ob- 
jectively made it necessary to use consultants. He pointed out that 
all data processing systems would require evaluation on each of the 
computer systems they might eventually consider for Marxson. He 
passed around copies of a talk given to Marxson executives by a 
systems consultant. This illustrated the detailed nature of the analy- 
sis of existing business systems which was required. A copy of this 
talk is included as Exhibit One. 

All members of the Committee agreed that extensive investiga- 
tion and systems design would not be done adequately by equip- 
ment manufacturers. They further concluded that the existing 
organization structure did not contain the technical manpower to 
undertake the systems design, although they had several possible 
candidates to work in such a group. Heilson discussed with depart- 
ment managers what personnel they might make available for his 
project. Out of fifteen potential employees, six candidates were 
selected after interviewing. These men were immediately enrolled 
in courses to supplement their existing knowledge and experience. 
They were encouraged to participate in professional meetings and 
university sponsored programs and courses. 

Marxson customarily worked very closely with his Treasurer, 
A. M. Biddle. Biddle was a member of several organizations which 
had conducted seminars on electronic data processing. He had at- 
tended these and been interested in this new tool because he recog- 
nized how valuable it might prove to Marxson. 

Early in December, at a regular meeting of Heilson's committee, 
Biddle presented data which he said represented the costs that 
would be sustained if Marxson and Company proposed to restudy 



196 JMarxson and Company 

their entire business data processing system. These costs had been 
obtained from other large industrial firms generally comparable in 
dollar volume and number of employees to Marxson. It appeared 
that it would require one and one half man-years per major area to 
be studied. Biddle stated there were at least five such areas in each 
Division of Marxson and Company. The staff costs approximated 
5200,000 on this basis, in addition to which there were contingent 
costs that might add as much as an additional 550,000 before one 
problem could be run on a computer. 

Biddle made other points from the experience of manufacturers 
who had taken up the challenge of using electronics in their business 
data processing. He showed examples of costly systems redesign 
made necessary by the selection of a highly specialized computer 
system. When the need for such redesign became necessary, it added 
to the already substantial costs. Biddle was also concerned about the 
problem of finding the right consultant for this task and of the 
additional costs they would sustain if they were to use outside 
services. 

One of Biddle's most impressive points was the competitive ad- 
vantage that would accrue to the first member of their industry to 
use a computer. It was obvious that the ability to process large quan- 
tities of data rapidly and react quickly to market changes and de- 
mands would give Marxson a decided advantage. Biddle emphasized 
that a complete system study would require at the very minimum, 
two to three years; thus, Marxson was a full four to five years from 
reaping the advantages of electronics. 

Biddle proposed that Marxson and Company immediately place 
an order for a medium scale computer from a well recognized equip- 
ment manufacturer and save a substantial period of time. Since this 
computer equipment had been used in a wide variety of business 
applications, it appeared that it would fit the needs of their firm. 
Biddle pointed out that studies could begin during the one year 
waiting period while equipment was on order, under the guidance 
of the selected computer manufacturer. With a medium scale com- 
puter, it was proposed that a few of the most urgent data processing 
problems could be prepared before the arrival of equipment; as the 
load increased, consideration could be given to a large or inter- 
mediate scale machine. 



Marxson and Company 197 

Biddle estimated that if payroll and other existing punched card 
applications were converted for the computer, only thirty per cent 
of its time would be taken on one shift, leaving plenty of time for the 
most urgent application, Inventory Control, which was being done 
manually. 

Heilson agreed there were some advantages to this method of 
getting started with computers. He presented some argument in 
defense of using the approach he had been exploring, but felt it 
unwise to argue this problem before investigating thoroughly to be 
certain that Biddle's point of view was not desirable. 

EXHIBIT ONE 
THE USE OF COMPUTERS IN BUSINESS OPERATIONS 

I am glad to be here today to talk to you about electronic data process- 
ing. I'm sure there is no subject you could have chosen that has more 
current interest or that will, eventually, have a greater effect on the 
operations of business. The interest of Marxson and Company in this 
subject is another indication of a great source of strength in our society 
— the willingness, even eagerness, to investigate new and better ways of 
doing a job. 

It is interesting to note that the first, large electronic computer was 
completed at the University of Pennsylvania at the close of World War 
II. This machine, the ENIAC, was the forerunner of all the machines 
which are now the hearts and brains of electronic data processing sys- 
tems. Today, there are over 2,000 electronic data processing machines 
in use by government and business. More than 200 of these are the large- 
scale variety, each representing an investment of more than $1,000,000. 
It is evident that business has been active in applying computers to bus- 
iness problems. 

Some day your company will be using electronic data processing 
equipment. The only question for me is "when?" The answer to this 
question is, "when the benefits of an electronic data processing system 
offset the cost of the installation." 

Do not assume I am building an unprovable or unfounded premise 
into this conclusion. I must be honest to tell you, however, in my objec- 
tive capacity, that: 

1. The operations of Marxson and Company, as I know them, in- 
dicate a good likelihood that you could be saving money by 
automating your data processing today. 



198 Marxson and Company 

2. The costs of business data processing are on the rise! It is only 
a question of time before all but the smallest companies take 
recourse to automation to handle their paperwork. In this regard, 
one of the things we sometimes overlook is that the trend in the 
cost of a manual or semi-automatic system is constantly increas- 
ing per unit operation. With the electronic system, an increased 
burden would mean simply more time on the equipment or, at 
some future point, more equipment. Even if this were the case, 
the increase in cost per unit operation would be substantially 
less than the corresponding increase in a manual or semi-auto- 
matic system. 

3. Your competitors are already exploring electronic data process- 
ing with the view of leasing a computer. 

4. Other companies have found it profitable to use computers. 

5. If there is any question, it is to determine the least costly and 
most effective way of making the transition to electronic data 
processing and in what areas to begin the transition first in order 
to maximize dollar savings. 

I would like to devote my talk today to this subject: What can Marx- 
son and Company do to determine how to use electronic data processing 
equipment in its own operations? 

The route to electronics must invariably begin with a system study, 
the purpose of which is to establish the economic and operational 
feasibility of converting existing manual or punched card methods to 
automation. 

The system study has four major components. We will see that they 
represent an orderly approach to solving your problems in the same way 
they have been used to solve complex commercial and military prob- 
lems. 

1 . The present system study: to determine the status of the existing 
systems, and other rules under which present business operations 
are organized. 

2. The present system cost study: to determine the displaceable and 
nondisplaceable structure of expenses that can be properly ac- 
crued to data processing, reflecting the present methods of opera- 
tion. 

3. The proposed system study: the proposed electronic data process- 
ing system which will do all — and sometimes much more — than 
its existing counterpart. 

4. The proposed system cost study: the anticipated costs of con- 



Marxson and Company 199 

verting, installing, and operating an electronic data processing 
system and the resulting short and long term costs and savings. 

A few words about each of these lour points will clarify the issue of how 
to make the transition to electronics. 

The way to embark on a present system study in your own company 
is to set up an executive steering committee whose function is to super- 
vise and steer the efforts of the men who will actually do the work. The 
steering committee should include members of top management and 
represent the major departments. 

A team of workers must be selected to make the study. It should 
include personnel and qualified consultants who specialize in electronic 
data processing. The probable make-up of such a group would be men 
from middle management levels representing the departments where 
the operational changes might be expected to be the greatest. The em- 
ployee members of the team should be assigned to work full time with 
consultant personnel, who would be mathematicians and engineers with 
specialties in business analysis, operations research, and electronic data 
processing systems. 

A planned educational program should be launched immediately in 
a two-pronged attack. First, there should be a program for Management, 
designed to explore as far as practical the various aspects of electronic 
data processing. Such a program, tailored to those who would not re- 
quire any knowledge of the technical problems, would concentrate on 
system engineering philosophy, cost benefits, and operational benefits 
of computers. This Management orientation type of approach is ex- 
tremely important. It is here, that Management and computer special- 
ists, achieve a mutual understanding of each others' aims and a tolerance 
for the day-to-day problems of converting to electronics. 

The second facet of the planned educational program should run 
concurrent with activity at your company. A nucleus of personnel should 
begin the process of learning about computers, utilizing as many of the 
front rank manufacturers' training schools as possible. A planned pro- 
gram would stagger this type of learning without putting undue accent 
on any one system of equipment. Such a program might continue for a 
year, during which time one to two months of schooling might be in 
order for each member of the team. More advanced courses can wait 
until the selection of equipment has been made and information can 
be specialized to the needs of this particular installation. 

The next step is an over-all survey of the company's operations. This 
should be a brief but critical examination of each area to be considered. 
As a result, you will arrive at a priority list based upon the need for 
mechanization and the susceptibility to mechanization. Rough measures 
for determining the need might be the number of personnel involved 



200 Marxson and Company 

or the cost of the present system. The susceptibility to mechanization 
depends upon the type of operation involved. Operations that are 
repetitive and performed on large volumes of data are particularly 
susceptible to successful mechanization and frequently show the largest 
savings. 

Another objective of the over-all survey is to delineate integration 
possibilities among the various applications. Even at this early stage, it 
is possible to visualize how the various data processing tasks may be 
interrelated to achieve the most economic and effective operation of the 
computer. 

Now you are ready to begin a detailed analysis of present system 
operations. The reports or documents that are produced by a system 
are called the outputs. The source documents going into the system are 
called the inputs. The outputs and inputs must be ingeniously inter- 
related in order to design an economic, electronic system. But to try 
this, without examining the present methods of operation, is to risk 
over-simplification — or failure to include some of the essential, re- 
quired processing. In a detailed examination of the present operations, 
it is possible to detect the unnecessary duplications of operations or 
files that typically creep into data processing systems that have evolved 
without scientific control. Elimination of such duplication is one means 
of effecting savings in a new system. A detailed description of the 
present system is a good starting point because it indicates one method 
of performing the required data processing operations. 

These are some of the ways to describe the present operations of your 
Company: 

1. Flow charts: illustrate the relationship of procedures or activities 
within departments or smaller groups. 

2. Samples of input and output forms and reports: provide a basis 
for examining how well or how much the information being 
transmitted or processed is being used or is necessary. 

3. Specification of the processing necessary to arrive at a required 
output from a given input: assures that all fundamental steps 
necessary to the proper completion of a task are going to be 
included in the design of a new electronic system. 

4. A list of the files to be maintained: assembles the data essential 
to the system such as the number and size of records, or the aver- 
age number of numeric and alphanumeric characters that com- 
prise any individual record — in electronic data processing nu- 
meric quantities can be stored in less space than alphanumeric 
quantities. Another aspect of files which must be specified, is the 
access requirement. That is, how many times, over what period, 



Marxson and Company 201 

is it necessary to refer to a file? In what sequences do the refer- 
ences take place? 

5. In addition, there must be a specification of tlie activity and 
other pertinent volume figures: For example, how many orders 
are there per day, or how many receipts? Again we must specify 
both the average and the maximum. In many cases, it is useful 
to make a distribution of the activity which specifies the fre- 
quency of occurrence. 

6. Any special communication requirements must also be docu- 
mented. 

7. TJie problem of centralization versus decentralization must be 
carefully isolated: Certain business operations may only be 
partially susceptible to a centralized data processing system. The 
need to have certain records available for interrogation twenty- 
four hours a day may defeat the orderly schedule of an electronic 
data processing center. 

The analysis of present system costs will naturally flow from the work 
which has been performed to analyze the system operations. The per- 
tinent elements to isolate are: 

1. Actual personnel costs — by department and section — devoted 
to the current processing of data. 

2. The associated costs of space, equipment, and other facilities 
necessary to implement the use of personnel. 

3. The costs of existing punched card installations, including 
equipment rentals and facilities. 

4. The costs of data processing by common tasks such as payroll, 
personnel accounting, cost accounting, production scheduling, 
inventory, accounts payable, receivable, and so on. These are so- 
called horizontal costs because they affect the operation of all or 
most departments. 

5. Opposed to these are the vertical or strictly non-common tasks 
that are associated with highly specialized data processing activi- 
ties, some of which may be suitable to centralized data process- 
ing. 

Analysis of the present system costs will enable your data processing 
task force to pick the areas in which the electronic system design will 
begin. Three or four areas only, might be sufficient to prove that elec- 
tronics will have a handsome payoff. At this point a priority of effort can 



202 Marxson and Company 

be established, and the task force can go to work to postulate — that is, 
design — the proposed electronic system. 

The same tools used to describe the existing system are re-sharpened 
as we go into the back stretch! The order of the day is flow charting 
and some additional technical work called block diagramming. The 
techniques describe the new way in which the data processing opera- 
tions can be done, bearing in mind some generalized electronic system. 

Frequent conferences with the steering committee assure the respon- 
sible persons in your company of the step-by-step progress of the study. 
The steering committee may meet once every six weeks — or more fre- 
quently, if necessary. As the new system begins to take shape, meetings 
generally become more frequent. 

But parallel with these reporting meetings are the many contacts 
your task force has with the operating members of your business to 
develop the new systems concept. These meetings prove the value of 
new ideas which are being introduced — so there will be a general con- 
currence of opinion regarding innovations that may be forthcoming. 

During this period, the philosophy of systems engineering — some- 
thing your employees must learn — is stressed. What makes a system? 
What are the system components? What attributes does a successful 
system have? What is integration, and what are the boundary conditions 
and subsystems that must be created to implement the design of a suc- 
cessful electronic data processing installation? 

These and many more questions become the preoccupation of the 
task force. The result, however, makes itself known in a superior, eco- 
nomic utilization of your personnel, facilities, and dollars. 

Side by side, the proposed system costs and proposed system design 
take form. The operational advantages will be paralleled by an analysis 
of the economic advantages. These will be assembled in two forms: 

1. A report to the steering committee containing an outline of the 
next step or series of steps which eventually will place a com- 
puter in operation on your problems. 

2. A request for bid suitable for analysis by manufacturers of equip- 
ment. Manufacturers, in turn, will prepare bids which must be 
analyzed by your personnel and your consultants. Their bids will 
be based upon their appraisal of how well their equipment can 
handle the system as proposed by your employees and consult- 
ants. 

By examining the equipment evaluations of each manufacturer, based 
upon the common postulated system, it will be possible for the consult- 
ant to perform the last step in the system study, to make conclusions 
and recommendations. Among these will be the economic evaluation 



Marxson and Company 203 

that must show the dollar advantages or disadvantages of electronic 
data processing. The economic evaluation is a comparison of the dis- 
placeable cost of the present system with the cost of the proposed elec- 
tronic system. There must also be careful consideration of the so-called 
"intangible" benefits — more rapid reporting, more accurate informa- 
tion, and more sophisticated statistics and reports — these will all have 
a tangible benefit to which a dollar value can be ascribed, even though 
in some cases it may be very difficult to do this. 

But we are not yet on the home stretch! Once equipment is on order 
the task force must begin the chore of converting human language to 
machine language. This process may take upwards of a year — during 
which time additional training, instruction, and formalizing of your 
electronics group must take place. In short, this waiting time for the 
computer to arrive is not wasted — far from it. Your employees will no 
doubt have as difficult a job before them in the home stretch as they 
did leaving the starting line. 

A word of caution — if you were to start your system study for elec- 
tronics tomorrow, depending on the size of staff and the amount of stand- 
ardization in procedures, you could assume it would require about a 
year to send bids out to manufacturers. This means that electronics for 
Marxson and Company is always two years away, at the very least! This is 
a long time to wait in the critical days of ascending costs and high budg- 
ets. 

Today you have heard how a typical electronic data processing in- 
vestigation might be performed. This was called a systems study. You 
heard about the necessity for an education program as an essential ele- 
ment in data processing plans. We then reviewed how the system study 
is performed, outlining important steps in its conduct. These steps in- 
cluded education, the over-all survey, a detailed description of the 
present operations, a postulation of a generalized electronic data 
processing system, a specific equipment evaluation, and, finally, the 
preparation of conclusions and recommendations. 

I commend to you these principles and the high quality results they 
have achieved for others, and can achieve for you, in planning the 
electronic data processing system for Marxson and Company. 

QUESTIONS 

1. What is the problem? 

2. Should this company undertake a complete business systems 
study preceding the choice of a computer? 

3. What would you do if you were Marxson? 

4. What problems are posed by the systems approach? 



13 



The International 
Corporation 



Fred Fox, the manager in charge of special projects, reported to the 
vice-president in charge of operations, at the national headquarters 
of the International Corporation, in Chicago. His staff was drawn 
from various departments of the home office, on semi-permanent 
assignment. In their regular work in outlying divisions of the cor- 
poration, these men, who were on loan, operated as consultants. 
Because of years of service and specialization in limited fields, they 
were looked upon as experts when they were sent out on special 
assignments. 

When a project was organized in another division, it was cus- 
tomary to select a project leader. The choice of a project leader 
sometimes had more to do with ability to maintain good client-con- 
sultant relations, than skills. Project leaders reported to Fox, di- 
rectly. It occasionally happened that a man would be assigned to 
more than one project, working as a leader on one assignment and a 
non-leader on another. Customarily, the project leader reported his 
progress to the Division management, as well as Fox. On most 
projects, especially those exceeding three months in duration, a 
steering committee was formed. The project leader was a member 
of this committee, together with representatives of division manage- 
ment and interested departments. Members of such a working 
group, called a task force, had no direct contact with division man- 
agement and very little direct contact with Fox. Division Managers 
were vice presidents on the same organization level as the officer to 
whom Fox reported. 

Three projects were in process when Fox took over this activity 
on March 1, 1956. One of these at the Dallas division was a reorgani- 

204 






The International Corporation 205 

zation of the sales department. This project was headed by a former 
member of the national sales staff. It had been in process for two 
months and its status was satisfactory, according to the Dallas divi- 
sion manager. The second project was at the Newark division and 
had just begun. This assignment was to reorganize the division's 
accounting activity. The project was headed by a former controller 
out of the Chicago office. The third project was at the Denver divi- 
sion and had been in progress for two years — since March of 1954. 
Fox began a routine investigation of the status of this project. He 
started by searching the files to determine the original motivation 
to begin this study. He then determined the exact nature of the 
assignment, the personnel assigned to the task, and the schedule of 
completion. 

Fox noted that the original assignment had apparently been put 
aside and a somewhat different assignment begun in July, 1955. 
Progress reports indicated that the original assignment was to de- 
termine the data processing requirements of the Denver division. 
However, soon after the project was started, two of the areas under 
investigation commanded more and more of the efforts of the task 
force. By November of 1955, the full efforts of the data process- 
ing personnel were being expended in the production control area. 
From March 1954, when the project was launched, until March 
1956 when Fox took over, there had been several project leaders and 
steady growth in the size of the task force, from two to six men. 

Soon after Fox took over the leadership of his department, the 
International corporate committee called for a review and formal 
presentation of the data processing study under way at the Denver 
division. The date was set for April 15th. It was clear that a com- 
mitment of six men costing over $90,000 per year required review 
and evaluation. Fox was unable to determine if the Denver division 
manager had asked for a review of this project because of his dis- 
satisfaction with its progress. Since he had no basis for asking for an 
extension, Fox decided that an immediate, first hand investigation 
must be made. 

Fox disqualified himself for this task, because of many urgent 
home office commitments. Since he had no regular staff to call on, 
he borrowed a member of the Tabulating Department, James 
Bunkker, to make the trip to Denver and report the results of his 



206 The International Corporation 

investigation. Bunkker reported to Fox three days prior to his de- 
parture, and spent this time familiarizing himself with the problem. 
Fox called the project leader in Denver to advise him of Bunkker's 
forthcoming trip. He stated the purpose of the investigation by ex- 
plaining his problem of preparing a presentation for the corporate 
committee. Bunkker was told to prepare himself to face problems 
of morale and leadership, as well as technical problems. Bunkker 
talked extensively to Fox in order to understand what Fox wanted 
him to do in Denver. They decided the presentation to the cor- 
porate committee would have to emphasize the outstanding features 
of the new production control system. They agreed to stipulate that 
it was necessary to finish this study before the feasibility study could 
proceed. 

Bunkker was disturbed by the wide variety of questions which 
needed answering. Although essentially unfamiliar with produc- 
tion control except as he had come into contact with these systems 
in his tabulating work, he felt he could probably evaluate the prob- 
lems with sufficient advance preparation. Accordingly he set about 
outlining what he would do during his one week of investigation 
at Denver. This outline is on the following pages. 



OUTLINE OF ACTIVITY 
FOR TRIP TO DENVER DIVISION 

1. Orientation: 

A. Meet with project leader. 

B. Explain the assignment. 

C. Meet Denver operating (factory) personnel. 

D. Tour physical facilities. 

E. Meet 2nd and 3rd shift personnel. 

II. Look at the present system and proposed revisions: 

A. What are the inputs and how are they used? 

1. The forms: who originates; their flow destination, useful- 
ness, duplication? 

2. Information: any unavailable; everything required? 

3. What part does tabulating play? 

B. What controls keep the system in balance and operating? 

1. Where are they; do they work; are there enough of them? 

2. Is everyone working with the same set of ground rules? 



The International Corporation 207 

C. Examine the outputs. 

1. What form are they in? 

2. Is all of the information in report form? 

3. Is all of the information necessary to report? 

4. Is there redundance in content? 

5. Are the right people getting the right outputs? 

a. Do they know how to act on them? 

b. Do they act on them? 

c. What is their reliability? 

d. Is there too much detail? 

D. Examine the feedback plan. 

1. What is the level of accuracy? 

2. How is the system affected by the level of accuracy? 

3. What are the feedback documents? 

a. Will they be timely? 

b. How many; of what type? 

c. Who is responsible? 

d. Will the feedbacks keep the system current? 

E. Make flow charts of 

1. The gross system 

2. The major sub-systems 

3. Plan to check 

a. Areas where changes have been suggested. 

b. Areas where changes are in order, but are not planned. 

III. Work on the operating level: 

A. Foremen or supervisors of operating departments: 

1. What are their duties? 

2. Scope of activity? 

3. From whom do they get information, and in what form? 

4. Contacts with other personnel. 

B. Assigned indirect labor (dispatcher, parts movers, timekeepers, 
inspectors): 

1. Respective duties of each. 

2. Relative authority and responsibility. 

3. Are they document originators, processors, handlers, or do 
they "hold?" 

4. Who gives them their orders; to what extent are they under 
the department foremen? 

5. What is the relationship of Material Review, Quality Con- 
trol and Factory Management? 

C. Watch the operations for 

1. Transit time between operations. 

2. Number of jobs in front of each machine. 



208 The International Corporation 

3. What are the frequency and types of upsets to normal, pre- 
planned operation? 

4. Is there a "shift" problem? 

5. Is the conception of how jobs start and stop accurate? 

6. Any obvious effects of variable lot size? 

7. Are count, time, and charge reasonably accurate? 

a. What does the operator do to write up his job card? 

b. Could it be done another way? 
D. Watch outside contracted operations. 

1. Do they plan; how far ahead; how? 

2. How much capacity and versatility do they have? 

3. Transportation problems in and out. 

4. Liaison problems and delays of what type? 

IV. Meet with Industrial Engineering 

A. Their organizational position. 

B. What programs are they working on? 

C. How do they see the proposed system? 

D. How do they plan to contribute? 

V. Meet with project leader — review the system as seen in investiga- 
tion. 

A. Areas requiring more analysis. 

B. Areas of disagreement. 

C. Pursue facts to clear up problems. 

D. Determine policies in problem areas. 

E. Restate the system. 

VI. Set up the system for installation: 

A. Implementation areas. 

B. Design and study areas. 

VII. Write report to Fox. 



QUESTIONS 



1. What is the problem? 

2. What is the organization problem at Denver? 

3. Are some pertinent pieces missing from the outline? Should 
some parts of the outline be deleted? 

4. How would you have put the systems concept to work in this 
investigation? 



14 



Carlysle, Inc 



On January 1st, L. James Carlysle became chairman of the board of 
directors of Carlysle, Inc. His successor to the active direction of the 
business was Henry Armstrong, former general manager of W. S. G. 
Manufacturing Industries, one of the country's most profitable and 
cost-minded organizations. Carlysle's organization chart (Figure 
14-1) is shown as Exhibit One. The sales position of Carlysle was 
considered better than ever. Volume had grown from §350,000 a 
year to $2,500,000 in less than six years. The backlog stood at $680,- 
000, and profits were at an all time high. 

Shortly after taking over, Armstrong asked for a customer-volume 
analysis of sales. This analysis is included as Exhibit Two. In an 
effort to learn something of the principal customers, Armstrong 
asked the sales manager, Wendell Phillips, to prepare a short sum- 



EXHIBIT ONE 

ORGANIZATION CHART 
CARLYSLE, INC. 



BOARD OF 
DIRECTORS 



L.J. CARLYSLE, 
CHAIRMAN 



H. R. ARMSTRONG 



SALES 
MANAGER 



L.W. PHILLIPS 



CHIEF 
ENGINEER 



R. C. WEST 



CHIEF 
METALLURGIST 



A. A. RANKIN 



MANAGER OF 
PROD. CONTROL 



CHIEF 
ACCOUNTANT 



S. W. SWENSON 



Fig. 14-1. 

209 



210 Carlysle, Inc. 

mary of the twenty whose volume contributed most heavily to the 
total sales and backlog. On fifteen of these, Armstrong asked Car- 
lysle's bank to do a thorough investigation, since accounts receivable 
were in five figures. 

A series of meetings were initiated by Armstrong as a means of 









EXHIBIT 


2 












CUSTOMER-VOLUME ANALYSIS OF SALES, 












CARLYSLE, 


INC. 








Type 






Receiv- 




Number 


Unfilled 


Age of 


. ° f 






ables over 




of 


orders 


oldest 


indus- 


Cus- 


Acounts 


30 days 


Years 


items 


on 


order in 


try* 


tomer 


receivable 


old 


sold 


sold 


hand 


months 


5 


A 


$ 9,000 


.$ 6,000 


2 


1 


S 40,700 


4 


2 


B 


500 


100 


1 


1 


2,000 


2 


1 


C 


61,200 


23,600 


2 


8 


88,800 


4 


3 


n 


7,200 


4,800 


1 


6 


3,000 


3 


1 


E 


23,400 


8,000 


5 


12 


18,000 


3 


4 


F 


7,800 


5,100 


2 


8 


25,900 


2 


4 


G 


89,100 


64,800 


4 


23 


11,600 


3 


1 


H 


1,700 


1,200 


4 


3 


20,400 


2 


4 


I 


12,400 


8,000 


5 


16 








2 


J 


1,100 


100 


5 


2 


1,000 


1 


2 


K 


5,400 


4,000 


3 


6 


171,600 


3 


3 


L 


48,500 


30,500 


2 


15 


90,800 


4 


3 


M 


1,200 


1,000 


4 


6 


3,400 


2 


5 


N 


18,600 


12,000 


4 


8 


48,000 


5 


1 


O 


5,700 


3,000 


5 


3 


12,800 


2 


5 


P 


400 





3 


7 








2 


Q 


12,100 


7,000 


2 


3 


25.600 


3 


1 


R 


3,600 


2,800 


3 


1 


4,800 


3 


4 


S 


1,100 


1,000 


3 


3 


2,100 


1 


5 


T 


350 


350 


5 


1 


800 


1 




\\\ others 


100,000 
$4 10,350 


65,000 
$248,350 






108,800 
S679.300 


7 


1, Mil 


itary electronics; 2, Aircraft; 3, Machi 


ne tools 


: 4, Instrumentation; 


5. Other. 



beginning his orientation to the current operations. Armstrong felt 
the need to meet his associates on a personal basis as soon as possible. 
but was satisfied at the outset to meet them as a group. In the first 
meeting, Armstrong asked the question, "How do you account for 
the unusually high profit rate?" 

Mr. Phillips, the sales manager, volunteered this viewpoint: 

You are aware that this is a precarious business. There are only a dozen na- 
tionally successful manufacturers in this field today. As a result, we take a 
large profit because the technology that enables us to survive is not generally 
obtainable elsewhere. Our prices are high because we take on a class of jobs 
that our competitors won't. We ask a high price because we are in a non- 



Carlysle, Inc. 211 

competitive market where our customers normally ask for small quantities, 
high quality, and fast delivery. Furthermore, our customers know these prices 
are high and are glad to get the merchandise. 

Mr. West, the chief engineer, took the position that "high prices 
or not, the outstanding feature of the Carlysle operation was their 
ingenious tooling." Subsequent discussion brought out how he and 
his staff had solved challenging problems that opened new markets 
to Carlysle. West pointed out that a sizeable part of the total volume 
was with companies who bought very complex, high priced castings 
which other manufacturers were unable to produce. Molds were 
procured from mold makers who specialized in this work. 

Mr. Rankin, the chief metallurgist, said that in his mind the prob- 
lem was only partly stated by emphasizing tooling and engineering. 
He pointed out how shrinkage or expansion, warpage, hardness, 
chemical composition, and tolerances were all dependent on a set 
of metallurgical factors. The selection of pouring temperatures, 
catalysts and introduction of control elements in the melt were also 
quoted as purely metallurgical contributions. Rankin said that 
these things might only influence profit indirectly, but this tech- 
nology made the difference between successful and unsuccessful pro- 
duction. From his standpoint, therefore, high profit was due in 
many respects to advanced metallurgical technology. 

Mr. Blue, the production control manager, made the comment 
that while "the boys are squabbling over details, they should all 
recognize that planning, scheduling, and meeting customers com- 
mitments on time contributed to a reputation for dependability." 
He cited the innovations that were responsible for increasing pro- 
duction and reducing lead times. Blue was unable to accept the phi- 
losophy of West and Rankin in total. He said experimenting and 
development — typical of processing most orders — built delays into 
schedules. These delays were sometimes unexplainable and created 
bad relations and antagonism between the customer and Carlysle's 
Production Control. He cited the hit or miss "genius-type" effort 
which was required to maintain the "Carlysle reputation." 

When Armstrong asked Mr. Swenson, the chief accountant, for 
his opinion, he could only say that the bidding and quotation pro- 
cedure administered by the sales department must be acknowledged 
as the fundamental reason for high profit. He said that Carlysle 



212 Carlysle, Inc. 

used a job cost system and costs were collected on each order as it 
went through the shop. Each worker reported his time against job 
order numbers that were posted in the shop. Each month he sub- 
mitted a summary of total labor dollars charged to job orders. There 
was a great deal of unexpected fluctuation in these reports, but gross 
figures pretty well substantiated what the key men expected. Materi- 
als were charged in on a standard basis and showed smaller fluctua- 
tion. Swenson had nothing more to add about high profits except to 
note that Phillips was an outstanding salesman. 

Armstrong began a series of individual meetings to explore fully 
the operations of each segment of the business. These first discus- 
sions were with Swenson. A review of the Monthly Operating Re- 
ports illustrated a long list of jobs in process, some of which had a 
mean cycle time four times as great as the average. In addition, a 
large number of jobs were unprofitable, while a small number were 
exceptionally profitable. A partial list of jobs is included as Exhibit 
Three. Armstrong was told the average selling price per casting was 
close to $6.00. Large excesses in material cost were laid to the need to 
cast orders more than once because the original lot would not pro- 
duce minimum shipping quantities. Excesses in labor were not 
clearly identifiable with cause. Costs were not distributed by depart- 
ment, cost center, or operator. 

The system of reporting was submitted to scrutiny. Labor and 
material dollars forecast were found to depend on original estimat- 
ing and bidding procedures. Such estimates were made by sales 
personnel who had with long experience been able to predict ac- 
curately cost and profit, according to Swenson. Although Swenson 
admitted there was considerable fluctuation in individual costs, he 
pointed out that over-all figures were amazingly close to sales projec- 
tions. Engineering was consulted, as necessary, by Sales, in the quo- 
tation process. 

Armstrong had several meetings with Phillips and members of 
his sales staff. These meetings revealed divergent points of view 
regarding what kind of business was good for Carlysle. Phillips was 
set against large volume contracts, arguing that the Carlysle method 
of operation was not suitable for low priced merchandise. Some 
salesmen, however, contested this attitude because they felt there 
were big sales possibilities in repetitive, large quantity production. 



Carlysle, Inc. 213 



EXHIBIT 3 





PARTIAL MONTHLY OPERATING REPORT OF SHIPMENTS 










MONTH OF 


OCTOBER 














CARLYSLE, INC. 




















Actual 


Actual 






Number 


Number 


Labor 


Material 


Labor 


Material 


Job 


Date 


Pieces 


Pieces on 


Dollars 


Dollars 


Dollars 


Dollars 


Number 


Issued 


Cast 


Order 


Forecast 


Forecast 


Reported 


Reported 


16761 


8/51 


304 


240 


530 


270 


870 


570 


16804 


10/51 


260 


180 


420 


180 


588 


290 


17110 


12/51 


871 


550 


1010 


460 


705 


421 


17867 


2/52 


602 


400 


900 


440 


611 


400 


20411 


4/52 


104 


81 


190 


90 


421 


177 


26711 


7/52 


114 


90 


220 


80 


80 


91 


26713 


7/52 


90 


60 


80 


20 


31 


21 


26745 


7/52 


62 


54 


60 


30 


42 


21 


26751 


7/52 


145 


110 


120 


60 


401 


162 


26867 


8/52 


600 


420 


480 


220 


320 


208 


26868 


8/52 


314 


220 


250 


120 


251 


124 


26869 


8/52 


224 


175 


210 


80 


304 


178 


26873 


8/52 


187 


160 


180 


90 


120 


45 


26888 


8/52 


200 


170 


190 


90 


160 


80 


26891 


8/52 


161 


125 


140 


60 


151 


63 


26894 


9/52 


145 


125 


130 


70 


110 


65 


26896 


9/52 


40 


20 


45 


20 


30 


8 


26897 


9/52 


180 


140 


310 


160 


280 


140 


26901 


9/52 


90 


40 


100 


45 


61 


20 


26902 


9/52 


1200 


675 


1500 


750 


1810 


620 


26903 


9/52 


1000 


850 


1890 


940 


1460 


800 


26904 


9/52 


188 


150 


400 


100 


421 


27 


26905 


9/52 


221 


190 


480 


120 


321 


87 



They said, "Not all jobs would be so profitable that they would be 
able to carry high overhead rates." They further argued that diversi- 
fication of jobs and higher volume jobs would be an insurance 
policy to Carlysle, which was, at present, entirely dependent on low 
volume-high profit items. 

Phillips was critical of the cost system. He said it was impossible 
for him to know what jobs were profitable or what jobs were un- 
healthy for Carlysle, because reporting was so inconsistent. He said 
further, he would never be able to prove what kinds of jobs he 
should be getting because, "Engineering, Metallurgy and Produc- 
tion Control can't settle on a good set of ground rules for running 
first articles." Armstrong found out that first articles were customer 
samples and always had top priority. 

Discussions with Blue gave Armstrong another side of the pic- 
ture. Blue pointed out that the Sales Department handed out de- 



214 Carlysle, Inc. 

livery ultimatums based on their promises, without consulting him 
first. He illustrated with data how commitments made to customers 
ignored his already overloaded schedule. He also pointed out that 
delivery dates were dictated by customers with little recognition of 
contingencies that always arose in the production of first articles. 
Blue said that changes in schedule, which were frequent, made it 
extremely difficult to operate economically with small quantity pro- 
duction. Blue asked Armstrong, "Do you think we should continue 
to run jobs on which we have never been able to make a profit"? 

West commented to Armstrong that, somehow, parts were never 
really manufactured to the prescribed engineering standard and 
that if the shop ever started doing this, they would have a lot 
less trouble. West indicated that the complex technology of manu- 
facturing would never get simpler, and that better controls on qual- 
ity were necessary. Some jobs were necessarily going to be losing jobs 
because the role of Carlysle was to take on the tough ones. He said 
he was certain that this was how they had made their reputation 
and built the business. A description of the investment casting proc- 
ess is included as Exhibit Four. 

Phillips asked each member of his staff to make recommendations 
he could incorporate into a Master Plan. Swenson submitted a re- 
port recommending an overhaul in all phases of the dollars and 
cents side of the business. West asked for a program of quality con- 
trol. Phillips asked for better bidding, quotation, and production 
scheduling procedures, so he could get more business and satisfy 
more customers. Blue asked for a plan to revamp the organization 
and responsibilities of the key people in the business, making the 
hub of the internal operation in Production Control. Rankin sub- 
mitted the shortest request, asking for an appropriation of SI 6.000 
to set up a metallurgical lab. 

exhibit 4 

description of the 

investment casting process 

at carlysle, inc. 

Investment casting is a technique oi casting that is many hundreds of 
years old. It was used successfully in the creation of art objects, but 



Carlysle, Inc. 215 

until the 1930s it had little technological development for industrial 
uses. In 1940, industry required high tolerance castings with finished 
surfaces that were difficult or impossible to machine. These parts, made 
from new steels, increased manufacturing costs. Some parts could not be 
successfully machined within required tolerances, because of hardness 
and other technical problems. 

Investment casting processes are very complex. Each part requires a 
mold which may have one or more cavities. The costs of these molds are 
high because cavities must be machined to close tolerances. Parts to be 
made in low volume customarily have one cavity. Parts to be made in 
higher volume may have more. Cavities must be machined in negative, 
so impressions made of these parts can be removed in positive. The 
process of machining the cavities is a highly skilled project, requiring 
wide tooling experience and frequently requiring extensive rework or 
scrap due to miscalculations or corrections. 

Gates are attached to the impression to facilitate pouring of metals. 
Gates are avenues through which metal will travel, to be removed later, 
after casting and cooling. Gates are machined into the mold at the same 
time the cavities are made. Gating must frequently be corrected as a 
result of poor impressions or misfills in the casting process. Molds range 
in size from 1" x 1" x 2" for a small example, or may go up to 8" x 16"D 
for a large example. They weigh up to 50 pounds and may cost as little 
as $200 or be as expensive as $10,000. Molds are made of aluminum or 
steel. The shop is operated on two procedures, one for first articles and 
one for production. The only difference between these is that first 
articles are in sample quantities of two or three pieces, while production 
is in larger quantities. 

The first article procedure begins with mold check-out. Molds are 
checked out by the engineering department. Injections are tried on pro- 
duction equipment to do this. The injection machine forces hot wax 
into the mold orifice under pressure. Wax travels through the main gate 
into the cavity and fills all subsidiary gates. Pressures of injection may 
vary from 60 to 400 pounds, and take from one-half to five minutes. One 
injection may suffice to reject or accept the mold, although it could take 
more than a hundred. 

After injection the molds are opened with knives or prongs, and the 
wax image of the part removed from the cavity with the gates. Any gates 
not necessary to the casting process are broken off and discarded in a 
scrap bucket, for reclamation. If no visual flaws are evident, waxes are 
taken to Quality Control for dimensional check-out. Any problems in 
injection or removal are noted on paperwork accompanying the mold, 
and selected pressure and holding time are recorded. Production quanti- 
ties do not go through wax inspection on a 100 per cent basis, although 
all waxes require cleaning. 



216 Carlysle, Inc. 

First article waxes are inspected for dimensional and surface condi- 
tion. Out of tolerance or otherwise defective waxes are cause for rejec- 
tion and molds must be sent back to moldmakers for rework. If molds 
are certified for production, the paper work is completed and the mold 
passed for payment and taken to stock. Any remarks regarding composi- 
tion of wax or indications of anticipated production problems are 
noted for use in production. Copies of this information go to the chief 
metallurgist. Molds are bought by customers on acceptance of first 
article. 

From this point forward, first articles and production have identical 
processing. When the production order is released, waxes are shot with 
an added factor to cover anticipated scrap. They are sent to the wax 
assembly department where they are set up prior to investing. In set-up, 
the wax department foreman follows the suggestion of the engineering 
department to insure proper gating. Despite the fact that gating is criti- 
cal, it is difficult to determine for sure how many parts per flask will be 
possible until the first article cast is completed. This type of data is 
customarily incorporated in production runs. There is, however, a limi- 
tation of no more than 13 pounds of steel or 30 pounds of aluminum per 
flask. 

Individual waxes are cleaned and partially inspected prior to set-up. 
Wax parts are mounted on pouring cups and specially injected gates. 
Parts and gates are welded into a unit called the sprue. If the set-up 
design is approved, set-up operators will assemble the sprue for invest- 
ing. Frequently, two or more sprues will be tried because of doubt as to 
the outcome of the melt when first articles are being tried. Sprues are 
marked as to direction of pouring for identification in the casting proc- 
ess. 

Setting up sprues is a tedious process. There are a variety of wax 
welding methods, and there is a tendency for sprues not to be uniform. 
There may be only two parts to a sprue or as many as two hundred, de- 
pending upon the size of the part and weight of metal to be poured. Wax 
sprues must be thoroughly cleaned and inspected prior to the next 
operation. 

The dipping process is the core of successful investment casting. In 
this operation, sprues are covered with a chemical solution. After dip- 
ping, the sprue is lightly sanded and left to dry. This operation is re- 
peated from two to five times, after which the sprue is wax welded to a 
steel plate. The sprue is now ready for investing. 

There are two methods of investing, one for ferrous, another for non- 
ferrous parts. In either case, a circular or square section of steel plate 
from 4" to 16" in height is sealed to the sprue. This square or circular 
section is called the flask. Flask sizes and heights vary widely with sprue 
size. Flasks are made of stainless steel and cost about $50 a unit. 



Carlysle, Inc. 217 

Investment is a fine plaster-like material that flows easily and dries 
with a fine surface. Flasks are filled with this cement-like material which 
is shaken down to eliminate bubbles and create a solid pack. This is left 
to dry naturally, and harden. The base plate is removed, after which the 
invested sprue is set upside down on a dewaxing cart and wheeled into 
a dewaxing oven. The dewaxing process takes eight to twelve hours. 
During this process the wax melts and drips out of the investment leav- 
ing a smooth, clean negative cavity of the entire sprue. Temperature for 
dewaxing is 225 degrees. From forty to one hundred and twenty flasks 
may be in the oven at any one time. 

Before the flask and its contents cool, it is transferred to one of five 
heating furnaces, with as few as 18 or as many as 72 flasks to an oven. 
Dewaxed flasks are mounted on top of each other using 1" x 1" x 1" 
ceramic blocks to create air space between each flask. Ovens are closed 
and furnace programs are set to bring temperatures up as high as 1800 
degrees at the rate of 100 degrees per hour, automatically. Heating cycles 
require up to sixteen hours and must be slowly lowered after a peak 
heat, to the required casting temperature. An oven load may contain 
flasks to be cast at several heats. When temperatures have been lowered, 
metal shot or bar is placed in melting pots and the metal is melted. Flasks 
are removed, red hot, from heating ovens and clamped on melting pots 
over the pouring hole. Flasks are evacuated and inverted as a unit so 
metal can fill the investment cavity. After pouring, the flask is removed 
and left to cool. 

After cooling, the metal sprue is knocked out of the flask, and the parts 
cut off the tree with an abrasive saw. Parts are then deburred, ground 
and finished for inspection and shipment. 

QUESTIONS 

1. What is the problem? 

2. If you were Armstrong, what would you do? 

3. What can you learn from the description of the investment cast- 
ing manufacturing process? 

4. What are the systems aspects of this case? 



15 



The Simpson-East 
Corporation 



The Simpson-East Corporation devoted its full efforts to research, 
development, and manufacture of advanced electronic equipment. 
It worked exclusively on military contracts with cost-plus-fixed- 
fee compensation rules up to 1954. Starting in business in 1940, as 
the result of a merger with one of the country's oldest names in the 
field of communications and electronics, Simpson-East developed 
a reputation for high quality. 

Early in its business life, Simpson-East began bidding success- 
fully on small quantity, highly complex electronics contracts. It 
was on contracts of this type that the corporation built its business 
volume. Profits were between five and six per cent. A system of 
project costing was used. 

In 1954, Simpson-East employed 500. There were 121 in various 
engineering activities. About 250 employees were direct labor. The 
balance were indirect, distributed over administrative and service 
departments. Dollar volume was $3,000,000 annually. 

The corporation had a wide source of business, among which 
were air-frame manufacturers as well as electronics firms. In addi- 
tion, they were invited to bid directly by various branches of the 
armed forces. Blueprints, written specifications, or samples were 
supplied to assist in bidding procedure. However, these contracts 
called out severe time schedules, since frequently a comparatively 
small part of a large electronics system could be holding up a size- 
able, urgent project. Consequently, it was in the interest of cus- 
tomer relations and sometimes programs of national importance, 
to reduce bidding time to a minimum. 

To cope with this problem, rules of thumb were conceived which 

218 



The Simpson-East Corporation 219 

made it possible to bid quickly. The motto was "get the job, and 
we'll figure out how to make money later." There was a uniformly 
low profit as a result of this method of operation. However, in 
renegotiation, profits were sometimes improved. 

A typical order for a new commodity would begin with a contract 
to do research, development, and manufacturing. The first desired 
output of the development phase was a circuit design called a 
"bread-board." However, work on the bread-board started before 
the end of the research phase. The final engineering design fol- 
lowed the completion of the bread-board. When engineering design 
was out of the way, a prototype was built. 

Production Control began its investigation of the new commod- 
ity "over-the-shoulders" of the research and development staff. The 
purpose of this early interest was to place on order hard-to-get 
items with long delivery times. This investigation period might 
range from 1 to 300 man-hours. During this period, Production 
Control was forced to accommodate its ordering procedure to the 
many changes in the design of the new commodity. There was a 
natural tendency to order as many of the parts as possible, and get 
them into stock. But this policy frequently led to contradictory 
situations. Frequently, expensive components ordered according 
to established rules were left unused in stock to complicate renego- 
tiation. This occurred because changes in commodity design could 
take place any time prior to delivery to the customer. If new com- 
ponents were required, the original components were sent to stock 
as surplus. 

An equally complex situation occasionally developed when com- 
ponents which should have been ordered, were not. Sometimes this 
problem was associated with lack of information at a sufficiently 
early date; other times it was laid to poor investigation or inade- 
quate record keeping and follow-up systems in Production Control. 
Production Control customarily requisitioned components. Order- 
ing and follow-up was done in the purchasing section of the mate- 
rials department. Materials that were ordered on CPFF contracts 
became the property of the government on receipt. The urgency 
to maintain minimum inventories was not as great as it might have 
been if these inventories required the working capital of Simpson- 
East. Surplus inventories were carefully watched by the government 



220 The Simpson-East Corporation 

and periodically reduced by direct sale to anyone who was in the 
market. 

All materials were placed on order for production following the 
completion of the prototype produced by the design engineering 
staff. Verbal or memo sales orders were the authority to purchase. 
There were several members of the engineering-sales-administra- 
tive management who could authorize the "go ahead" on a project. 
Material requirements were taken directly from blueprints by 
Production Control clerks. There were no parts lists or bills of 
material issued separately by Research, Development, or Design 
Engineering. As a consequence, liaison was required whenever 
new commodities were placed on order. One of the big problems 
was whether or not to use standard parts whenever possible. 
Descriptions of components were not standardized, with the result 
that frequent changes, revisions, and clarifications were necessary 
to meet the manufacturer's specifications. 

Production analysis was the next step in the procedure. In this 
period, individual parts and subassemblies were analyzed in antici- 
pation of actual production. Parts lists were indented at this time, 
and lead times were estimated based upon the best information 
available. Analysis always revealed inconsistencies and mistakes in 
design, which made further changes necessary so the product would 
be possible to manufacture. All changes in design or parts require- 
ments were processed through a Change Board that met once a 
month. 

The production phase followed material order by approximately 
30 to 45 days. The expected problems of tolerances not adding up 
and last minute revisions were accented by the embellishments to 
products in the process of their manufacture. There was no rule 
to apply to changes which were necessary versus changes which 
merely added to the engineering perfection of the product. The 
result was that production lines were continually in flux. Engineers 
were active in all phases of the actual manufacturing process. A 
bar chart (Figure 15-1) describing the phases of operations follows 
as Exhibit One. 

In February of 1956, the accounting department, with Manage- 
ment, reviewed the working papers for calendar year 1955 opera- 
tions. The here-to-fore acceptable profit had slipped from 5.4 per 



The Simpson-East Corporation 221 

cent to less than 2 per cent. Reduced profits were associated directly 
with fixed price contracts which had become an increasingly im- 
portant factor to provide dollar volume in early 1955. CPFF busi- 
ness, formerly 99 per cent of the total dollar volume of business, had 
slipped to 28 per cent. 

Management reacted with a shakeup in personnel. Several ex- 
perienced members of top management were replaced, including 
the general manager. New personnel were sought to fill the vacant 
posts. It was anticipated that some of the new personnel would 
come from the electronics industry; others would have diversified 
backgrounds. Joe Waddell, the new general manager, was among 
the latter. 

When Waddell took this position in May, 1956, he did so with 
the understanding that he had six months to begin to show some 
acceptable signs of progress. W'addell was not impressed with the 

EXHIBIT ONE 

TIME SCALE OF A TYPICAL ORDER 
SIMPSON -EAST CORP. 






1. 

2. 


Research and Development 
Bread-Board 








1 h-- 




1 


3. 


Design 


1 


— 1 


1 


4. 


Prototype 




i 1 




5. 


Investigation of Requirements 














6. 


Test 








7. 


Material Order 




!__, 


-I 


8. 


Production Analysis 












9. 
10. 


Manufacturing 

Final Inspection and Test 








1 1 1 1 1 


I l 1 i i i 


l l i i 



TIME IN MONTHS 

Solid lines are periods of initial activity. 

Dotted lines are periods of liaison to incorporate changes, etc. 

Fig. 15-1. 



222 The Simpson-East Corporation 

Simpson-East management, and he knew they were in trouble. But 
they had given him a free hand to do what he thought necessary 
in order to provide an acceptable level of profits. 

By June, Waddell was obtaining daily information on all phases 
of the Corporation's operations. In addition, he had accumulated 
a notebook of data, outlining the functions of various departments, 
and brief flow charts showing the interrelationships of the various 
groups within each department. A staff of two men assisted in the 
development of data, and pursued the leads and suggestions he 
made in order to obtain a clear picture of the existing mode of 
operation. 

Waddell's attention was called to a report submitted by a plant 
committee four months before he was installed as plant manager. 
Since many of the members of Management were impressed by this 
report, he gave its contents his careful attention. (This report 
follows as Exhibit Two.) 

On July 10th, Waddell advised Management he was ready to 
meet with them and submit his findings. 



exhibit two 

cost reduction: 
a plant committee study 

This report outlines what our committee considers to be the most 
fruitful areas of overhead cost reduction. 

The approach utilized is best described as The Group Process in Ad- 
ministration. In our earlier meetings, the problem was outlined in terms 
of what has happened in overhead cost and what we expect to happen, 
dollar-wise, during the remainder of 1955. Keying our investigations to 
the major cost factors, the indicated group members attacked the prob- 
lem from two points of view: (1) on a departmental basis; (2) on an 
integrated or "across-organizational lines" basis. 

What follows is a series of pointed "cost-cutting" recommendations 
which emerged from our integrated group discussions. These sugges- 
tions, if implemented, will result in significant reductions in operating 
costs. 

(1) Divisional planning of new business and project activities as 
they develop must give consideration to realistic lead times. 
Major economies can be realized by "building in" sufficient 






The Simpson-East Corporation 223 

lead times to provide for known, fixed time requirements. 
Among these are: 

a) Reduced material procurement costs 

1) Greater productivity realized by individual buyers 
through systematization of material procurement. 

2) Cost reduction on air freight expenditures. 

3) A lowering of clerical personnel requirements resulting 
from lesser complexity in incorporation change orders 
in purchase orders; special follow-up systems. 

4) A lowering of telephone costs in placing "crash" orders. 

b) Reduced overtime compensation in R 8c D and produc- 
tion activities. 

c) Reduced personnel recruiting costs in terms of "crash" 
advertising program, recruiting trips, and so on. 

d) Reduced labor costs by planning staff requirements rather 
than staffing to meet crash situations. 

(2) Establish a clear definition of project and manufacturing re- 
quirements and dissemination of these requirements on a 
need-to-know basis. Significant overhead cost reduction results 
in terms of a reduction in duplication of effort in production 
planning and in material procurement. 

(3) Accomplish a transferral of labor costs of significant numbers 
of employees from indirect to direct where industrial prece- 
dents in government contract activities have been established. 

(4) Initiate a training program which will result in significant 
"upgrading" of current employees, decreased costs of recruit- 
ing experienced people, and reduced turnover ratio. Suggested 
programs for current implementation include 

a) A formal "offhours" technician training program given by 
several members of the technical staff and professors from 
local colleges and universities. College credit should be 
given where possible. 

b) A similar program for junior engineers. 

c) An assembler training program. 

d) An apprentice machinist training program. 

e) A supervisory training program. 

(5) Install a central laboratory equipment control to project 
equipment requirements and utilize currently available equip- 
ment to an optimum level. This comprehensive control would 
result in major cost reduction in terms of capital equipment 
investment, overhead servicing, and depreciation. Suggested 
implementation is the installation of the function in the 
technical section. 



224 The Simpson-East Corporation 

(6) Standardize the nomenclature of components. All components 
ordered on purchase requisitions or specified on designs should 
be denned in a predetermined format providing for their com- 
plete description. The areas of economy would be as follows: 

a) Reduced cut in man-hours in both production and publica- 
tions groups by making possible a straight-forward, nearly 
"one-time" material analysis, minimizing the number of 
check and correct operations. 

b) Reduced costs in terms of purchasing man-hours spent in 
checking, correcting, and modifying information given to 
vendors. 

(7) Define a standard range of continuously used components. A 
significant result of standard nomenclature would be a set of 
standard parts composed primarily of hardware and inexpen- 
sive electronic components. Initiation of this program depends 
upon acceptance by project and design engineering. Resultant 
economies would be: 

a) Inventories can be reduced by elimination of duplications 
with resulting savings of space and personnel requirements. 

b) Large-quantity purchases, from prime sources rather than 
jobbers, could be made to cover all needs. Jobbers pre- 
miums in many cases are as high as 50 per cent of the pur- 
chase price. It is suggested that production stores act as 
supplier to the various stores, to capitalize on large 
quantity purchases and reduce inventories. 

c) A reduction of numbers of purchase orders with resultant 
reductions in overhead clerical activity. 

d) Paper work economies by use of "Buy Cards" in Produc- 
tion replacing purchase requisitions. 

(8) Establish a stock committee. This committee would meet once 
a month to consider anticipated component requirements, ad- 
just stock levels for parts which have not moved during the 
month, and act as a sounding board for functional problems 
related to stock and its issuance. Economies would be 

a) Reduction of dollars tied up in obsolete and slow-moving 
stock. 

b) Minimization of the necessity of expensive rush procure- 
ment of parts. 

c) Less clerical time spent routing complaints concerning the 
stock room to Management. 

(9) Eliminate duplication of material follow-up by Production 
and Purchasing. These operations presently follow a parallel 
course. Major economies would result from combining these 



The Simpson-East Corporation 225 

operations either as part of Purchasing located with Produc- 
tion or as part of Publication. 

a) Staff required could be reduced. 

b) Files could be combined and clerical personnel require- 
ments reduced. 

(10) Coordinate the cost accounting system between the operating 
section and the accounting section. A cost system developed in 
this manner will enable Production and R 8c D to reduce the 
amount of clerical effort currently required. 

(11) Install mechanical tabulating and record keeping equipment. 
The estimated cost of a tabulating section indicates that it can 
be operated at a lower cost than if our full tabulating services 
are provided from outside. An advantage of operating our own 
equipment is long-term costs reduction, not now being realized, 
which will accrue as a result of accounting data and reports 
being generated on time with less clerical effort. 

(12) Institute a change in material procurement procedures which 
will provide for charging more than one job per purchase 
order or purchase requisition to the same vendor. Cost reduc- 
tions will be in terms of reduced numbers of purchase requisi- 
tions, and consequent lessening of clerical effort. 

(13) Increase present personnel staff by the addition of two ex- 
perienced men. Reduced costs will result in two major areas: 

a) Employment effort can fully exploit the personal contact 
approach to recruiting, resulting in significant cost reduc- 
tions in terms of local advertising and extended recruiting 
trips. Contacts include present employees, educational 
institutions, employment agencies, and other companies. 
One local company reports that 60 per cent of new hires 
result from personal contacts. 

b) Employee turnover ratio can be reduced by closer screen- 
ing of new hires and an increase in the informal education 
program for supervisors in selection techniques. A reduced 
turnover ratio will result in significant savings in advertis- 
ing and recruiting costs. 

(14) Install procedures to insure more effective control of sick leave. 
Enlightened administration of the sick leave policy by super- 
visors will result in minimizing cost by abuses of this policy. 

(15) Use short form employment applications for initial screening 
purposes. This results in a lesser expenditure for the formal 
employment questionnaires. 

(16) Provide a less expensive recruiting brochure to be utilized in 



226 The Simpson-East Corporation 

non-professional recruiting. Significant savings can be realized 
by utilizing a more concise, less heavily illustrated booklet for 
non-professionals. 

(17) Adhere to the Temporary Added Compensation Policy for 
reimbursement of exempt personnel overtime. A significant 
reduction in premium time payments can be realized. 

Evaluation 

The committee feels that there have been several fruitful results of 
its group activity: 

First of these are the concrete recommendations for cost reductions 
which the committee believes to be sound. 

Second, a major value has been the free discussion of the operation of 
other sectors of the administrative activity in our plant. The resultant 
learning has been rapid and of permanent value. 

QUESTIONS 

1. What are the systems aspects of this case? 

2. If you were Waddell, what would you submit as your findings? 

3. What is the basic problem to which Waddell must address him- 
self? 

4. How do you evaluate the Plant Committee Report? 



J 



16 



Beaver Alliance Aircraft 
Corporation 



Beaver Aircraft Corporation, was one of the nation's small scale 
manufacturers of airplanes prior to 1940. Working on a job shop 
basis, Beaver had three models, all of which were considered highly 
reliable. Contracts with the Army were so small that there was no 
opportunity to enlarge the Corporation's activity, with the result 
that a small lot method of manufacture had been undertaken. 

In a monthly schedule of deliveries which might vary between 
eight and twelve planes, it was customary to make structural 
changes effective in odd months in order to minimize manufactur- 
ing problems. Since the backlog of orders was relatively small, job 
lots were seldom over one hundred pieces. The frequency of engi- 
neering changes was high, and, since airplanes were not identical in 
material or labor input, the need for accurate charging of direct 
labor costs had been recognized as a significant problem. 

Airplanes were assembled in large hangars. Crews were rotated 
across these locations, 19 in number, called work stations. Each 
station had a tool crib, parts crib, and a few fixtures to provide for 
access to the external parts of the plane. Mating operations as well 
as minor modifications took place in these stations. Station 20 was 
left vacant when the airplane was towed outside the hangar for 
checkout, engine run, and testing prior to its initial flight. As soon 
as station 20 was open, aircraft would be manually moved forward 
into the next position. 

With the advent of war in Europe, Beaver Aircraft foresaw that 
demands on their ability to produce would be so far in excess of 
their past capabilities that they began negotiations with Alliance 
Aircraft Producers, Inc., to effect a merger. Alliance had facilities 

227 



22S Beaver Alliance Aircraft Corporation 

and good production techniques, whereas Beaver felt they could 
supply the designs. The merger was endorsed by the government 
because of the need for large capacity. A method of operation was 
developed, and, in 1941, the Beaver Alliance Aircraft Corporation 
was formed. After physical consolidation in 1943, large scale pro- 
duction was begun, with schedules rising sharply month by month 
for several years. The plant with which we are concerned employed 
2,200 direct labor. 

Because of the nature of cost-plus contracts, the philosophy of 
laissez-faire was easy to come by. But alert management pointed out 
that contracts were subject to renegotiation and every detail of their 
cost structure would be subject to intensive scrutiny, especially 
direct labor. It was obvious that over-staffing the labor force would 
not operate to the disadvantage of the company, since there would 
theoretically be a return on every dollar expended, irrespective of 
the efficiency with which the end product was produced. The prob- 
lems of direct labor were classified as dual in nature. The first had 
to do with efficiency in production, the utilization of labor, and the 
minimization of cost to the government. The second area was that 
of accounting for the labor expended, whether efficiently or not, 
and seeing to it that work was properly charged, the time accurately 
stated, and the count of production consistent with lot releases. 
Methods of collecting costs varied between feeder shop operations 
and line assembly operations. 

The problem of efficiency of operations was in the province of 
Industrial Engineering. It fell on this department to set time stand- 
ards for each operation and compute the total standard labor hours 
required to complete each end product by division, department, 
and cost center. As a result of the activity of any period, usually a 
week, reports were issued comparing the actual performance with 
anticipated (standard) performance. Cost centers were said to be 
fifty per cent efficient if they produced fifty standard hours in a 
period when one hundred actual hours were consumed. The nota- 
tion to measure efficiency was 

Standard Hours _, „ ,. . /T7a: x /i\ 

= % Realization (Efficiency) (1) 

Actual Hours 



Beaver Alliance Aircraft Corporation 229 

Or conversely, the standards could be adjusted by the efficiency 
factor to change them into actuals. If the industrial engineers pre- 
dicted that a new or revised job would take a certain number of 
standard hours, the anticipated actual hours could be set up and 
labor requisitioned accordingly. The notation to show this adjust- 
ment was: 

Standard Hours . , TT /m 

. = Actual Hours (2) 

% Realization 

These relationships were expressed for purposes of labor loading 
in terms of hours. For accounting purposes, they were expressed as 
dollars. Sometimes learning curves were fitted to jobs where train- 
ing was an essential ingredient of efficiency. This type of system was 
extended into the field of direct labor forecasting in order to predict 
what levels of employment should be anticipated for various vol- 
umes. The shop average for efficiency was 60 per cent. 

The problem of sorting increments of time so that they would 
be charged to the proper job was the function of the Labor Dis- 
tribution Report. The operator recorded his time on daily Produc- 
tion Tickets indicating the airplane number, if the worker was on 
the production line. A job number that covered a certain series of 
airplanes was used if the operator was working in one of the feeder 
shops. The accuracy of labor distribution was considered of prime 
importance, although inconsistencies were apparent when charges 
to one series of airplanes fluctuated as opposed to the charges on a 
prior similar series. Occasionally a situation would arise where 
charges would accumulate for work that was not scheduled to take 
place on a certain series. The conclusion in this case was that the 
operator had made an incorrect charge. 

The feeder shops encountered a different type of situation. The 
count of production frequently varied from one machine operation 
to the next. Sometimes it was necessary to pull parts out of the 
productive sequence for purposes of scrap or material review. The 
subsequent rework to make bad parts reusable with the special one- 
time customer deviation, created a vast complex of paperwork. This 
made it difficult to associate a particular group of parts with their 
proper effectivity series. The vastness of the shops and the variety 



230 Beaver Alliance Aircraft Corporation 

of changes to complete a part prior to installation or subassembly 
created the necessity of lot identification. 

Lot identification in the manufacturing process was accom- 
plished by using a Travel Card, sometimes called a Traveler, or 
Factory Routing. These cards called out each operation, the depart- 
ment in which it would be performed, outside contracting opera- 
tions, and the standard hours per unit. This card was designed to 
stay with the parts throughout their processing life until they were 
sent to stock or until they lost their unique identity as a detail and 
went into a subassembly. At this time a new Traveler would be 
issued with the order to withdraw parts from stock, to start the 
process of assembly. 

The Production Ticket carried time data on each job. In the 
feeder shops, the operator wrote in his start time on a job when it 
was ready for setup. Setup time included everything from checking 
out tools to making the machine ready to operate, as well as produc- 
ing the first few pieces. The supervision in each department was 
charged with checking out the operator's setup and the good parts 
before calling for a first article inspection. The floor inspector 
would sign off the good parts or send bad parts to material review 
if they did not meet the minimum engineering requirements. If 
the parts were satisfactory, the operator would mark his ticket to 
indicate the setup had ended and the production sequence had 
begun. All setups were charged as indirect. Direct charges were 
classified by the controller of the corporation as "any operation or 
work that changes the size, shape, weight, appearance or usability 
of detail part, subassembly, or assembly." 

Indirect charges were — by the process of elimination — anything 
not direct, anything done by a nonproducing department, or any- 
thing not a part of the factory routing such as rework, repair, or 
overtime. Producing departments were those classified as using 
direct labor to accomplish a given amount of work. Indirect de- 
partments were looked upon as service departments, since it was 
argued that the service departments kept the direct or producing 
departments operating. Among the service departments were Main- 
tenance, Inspection, Production Control, and Transportation. 

When an operator in a feeder shop began a productive sequence, 
he did not have to complete his Production Ticket (Figure 16-1^: 



Beaver Alliance Aircraft Corporation 231 

this was done at the conclusion of the operation. The Traveler 
served to supply the needed information as to part number, job 
number, operation number. The operator might work on an opera- 
tion for only an hour or for several days, however the basic require- 
ment was no less than one card a day, since payroll was computed 
from the Production Ticket. Operators were responsible for the 
accuracy of Production Tickets. 

The clocking-out process on a job sometimes preceded the tear- 
down of the machine and sometimes followed it. No particular 
accent was placed upon the timeliness of this function since it was 
supposed to reflect only a very small increment of time. At the end 
of the day, the operator punched his ticket at a clock and put it in 
the "out" side of the rack under his employee number. 

Certain nonproductive time was not charged but assumed. Man- 
agement policy endorsed such activities as bond drives, Red Cross 
Blood Donation campaigns, and various other employee activities, 
as well as proper union activity during working hours. In addition, 
the Management policy allowed two five minute clean-up periods 
prior to lunch and quitting time and two ten minute rest periods, 
on each half of the shift. There was also a policy of allowing six 
minutes per hour fatigue and personal allowance for each em- 
ployee. Other factors that might keep machines from being oper- 
ated, such as absenteeism, power failure, machine breakdown, loan 
to other departments, or random delays had no place in preplanning 
and were likewise assumed. 






CLOCK 
RINGS 


REG. 
HRS. 


O.T. 
HRS. 


PART NUMBER 






OP. NO. 






LOT NO. 


O 

s 

O 

G 
l 


\ 






| 






\ 


MACH. CLASS 


MACH. NO. 


SPLIT LTR. 


g 

H 

c 

i 








I 


CLOCK NUMBER 


SHIFT 


HOME DEPT. 


DEPT. CHGD. 




BEAVER ALLIANCE AIRCRAFT CORP. 






} 






s 

H 

•z 

c 

1 

9 


o 

13 
W 
33 

1 
z 

c 

1 


PRODUCTION TICKET 






GOOD PARTS FINISHED 


JOB NUMBER 


TOT. ACT. HRS. 








START QUANTITY 


STD. HRS. /C 


TOT. STD. HRS. 


Total Actual Hrs. 



















Fig. 16-1. 



232 Beaver Alliance Aircraft Corporation 

Fourteen timekeepers working out of the accounting department 
had the function of totaling and extending a day's Production 
Tickets. They were stationed in the factory near the departments 
whose tickets they handled and were available if needed during the 
day. As a part of their routine, they checked the tickets of em- 
ployees not reporting for work and reported these personnel to the 
personnel department. The personnel department tried to deter- 
mine the reasons why an employee did not report for work. 

In October of 1944, the controller's office placed a request before 
Division Management regarding the conflict of interest in the end 
of the Production Ticket as an input to both Payroll Accounting 
and Cost Accounting. Mr. James L. King, chief of Cost Accounting, 
commented as follows: 

Our system of accounting control must be improved in several ways, although 
changes cannot and should not be accomplished too hurriedly. The account- 
ing department feels that the information on the Production Ticket is in- 
accurate. The quantity and quality of information both contribute to a 
variance of 30 to 40 per cent that we are unable to explain to the Government. 
The functions of the timekeeper must be changed. Some reconciliation be- 
tween total hours of work and hours spent on each job must be installed. 
Greater accuracy in counting parts must be obtained. Inspection count must 
be integrated into this system as a control factor. We should consider the use 
of pre-punched tabulating machine cards to be distributed to each job in 
order to minimize the amount of information necessary to be manually in- 
serted on the card. The control of labor in process at the present time is 
impossible, since the information being accumulated is not reliable. We need 
a system that will deliver cost data within 5 per cent accuracy. 

A committee was formed to study the problem. In the memo 
which directed the formation of this committee, Mr. Roland 
Welch, division manager, made this statement: 

Of the two principal inputs to the cost or value of our commodity, direct labor 
is more important since it is such a large percentage of the selling price. For 
this reason, accurate control over the distribution of time and proper nota- 
tion of the facts concerning the expenditure of direct labor are essential in 
our system. The committee set up to overhaul the methods for accruing direct 
labor is charged with creating an integrated system that will serve all parts 
of the operating business and fill all of its requirements. 

The first output of the new committee headed by Mr. King was a 
proposed timekeeping procedure. 



Beaver Alliance Aircraft Corporation 233 

PROPOSED FEEDER SHOP TIMEKEEPING PROCEDURE* 

1.0 Purpose: Establish uniform methods of 

a. Recording set-up, production, off-production, and tear-down 
worked on each operation on lots scheduled. 

b. Recording scrap and good pieces completed in each operation 
on each lot. 

c. Controlling work on lots according to schedule. 

d. Coordinating the supervisory, record-keeping, and expediting 
functions. 

e. Increasing accuracy of data on which effective machine loading 
and scheduling is based. 

2.0 All direct labor departments are affected 

3.0 Responsibilities 

3.1 Production Control 

a. Scheduling lot releases. 

b. Computation and release of raw material and castings. 

c. Preparing Manufacturing Releases, scheduling work each 
week. 

d. Keeping records of work-in-process. 

3.2 Dispatcher and Counter-Trucker 

a. Following and expediting work in all departments. 

b. Keeping current on status with timekeeper. 

c. Lot identification. 

d. Reviewing priority of jobs ahead with foreman. 

3.3 Foreman 

a. Review of lot schedules and job cards. Determines priority 
to each job. Assigns men and machine. 

b. Supervision of operators, checking their set-ups and parts 
counts. 

c. Arranging for inspection. 

3.4 Timekeeper (one per eighty direct labor employees) 

a. Assigning jobs to the operator on the machine or machines 
designated by the foreman. 

b. Recording all time reported by operators: 

1) On set-up and /or tear-down. 

2) On production. 

3) Off production. 

c. Maintaining up-to-the-minute recording of operations 
completed on each lot, including quantities of good and 
scrap parts. 

* Payroll accounting to be done separately through a weekly time card under this 
system. 



234 Beaver Alliance Aircraft Corporation 

d. Computation and reconciliation of times and quantities 
and balancing with time on weekly payroll cards. 

4.0 Shop Scheduling, Planning and Release of Lots 

4.1 Each Thursday, Production Control prepares raw material 
releases for lots to start running during the next week. 

4.2 The dispatcher and trucker arrange delivery of material by 
Monday. 

43 The timekeeper receives a Manufacturing Release showing 
the quantity of parts in each lot to be started during the cur- 
rent week from the dispatcher, after review by the foreman. 

5.0 Preparation of Job Cards (3 types) 

The Manufacturing Release shows all operations in sequence to 
be done on the part, the quantity to start, and the minimum quan- 
tity of good pieces needed in stock from the last operation (prime 
quantity). It shows the dates (Friday, week-ending date) each 
operation must be completed. From this, job priority is decided by 
the foreman and dispatcher. The timekeeper is interested, first, in 
those operations to be done in the cutoff department. For each 
operation, he must prepare one Setup Ticket. He does this as soon 
as he receives releases each week and always in advance of the first 
shift Monday. 

5.1 Set-up Ticket (Red printing) 

a. Timekeeper enters in two places on the Setup Ticket: 
part number; operation number; lot number (this is the 
same number for all parts released the same week), and 
clock number. 

b. Timekeeper enters in one place on the form: home de- 
partment; the department to be charged; shift; machine 
class; machine number; standard hours per hundred: start 
quantity; job number, and split letter (if any). 

5.2 Production Ticket {Black printing) Enter same information 
as set-up Ticket. 

6.0 Shop Schedule Board 

A series of job card racks will be arranged side by side on a vertical 
board in the timekeeper's enclosure. Each row of pockets across 
the board is for one of the production machines in the depart- 
ment. In the first pocket, cards will be placed for the job being 
set up or running on the machine. In the next pocket, the next 
job scheduled to run; in the last, other jobs ahead, in order as they 
should run. 

When the cards for each new lot have been completed, the 
timekeeper will review them with the foreman and dispatcher, if 
available at the time. The foreman will assign each job, indicating 



Beaver Alliance Aircraft Corporation 235 

the operator and a specific machine (or may omit this if any one 
of the machines of the machine class will do). Cards will be placed 
in the proper pockets, in line with the machine or group of 
machines assigned; or, arranged with other jobs ahead in the order 
requested by the foreman. The Set-up Ticket will always be on 
top. 

7.0 Timing the Tooling Tear-down and Set-up of New Jobs 

The foreman will follow each operator's progress on jobs and will 
determine with the timekeeper well in advance (two hours) the 
next job assignment, and will go to the tool crib and arrange for 
tooling to be ready. When an operator is ready for job assignment, 
he will come to the timekeeper, who finds the next job set-up card 
in the board, and tells the operator the part number, machine, 
and the number of pieces to start. 

The timekeeper will clock-in the set-up card and place it in 
the rack in the first pocket (Jobs Running) next to its machine 
number, with the Production Card behind it. He will record the 
date started on his copy of the Manufacturing Release. 

The operator will obtain tooling from the tool crib, tear down 
his last job, set up the new job, and run pieces for Inspection and 
foreman's OK. He will then report completion of set-up to the 
timekeeper who will clock out his Set-up Ticket. 

8.0 Timing Production 

The operator will notify the timekeeper when he is ready to start 
production. The Production Ticket will be clocked-in. Operator's 
clock number will be entered at two places on the card which is 
then filed in rack. It will be visible at all times while production 
is running. 

When the operation is finished, or at the end of the shift, the 
operator will count and report good pieces completed and scrap. 
Timekeeper will clock out and record pieces on the Production 
Ticket and if operation is completed for the entire lot, will record 
date completed, total pieces good and scrap on the Manufacturing 
Release. 

9.0 Timing Interrupted Set-ups or Production (Off-Production Ticket 
— Green printing) 

An Off-Production Ticket will be filled-in, in the same way as 
Set-Up or Production Tickets and clocked-in whenever an opera- 
tor reports work stoppage either on Set-Up or Production. In the 
Job Order box, enter the account number that explains the reason 
for the interruption. Also write the reason in the log column of 
the Manufacturing Release. 

When production or set-up begins again, the operator should 



236 Beaver Alliance Aircraft Corporation 

be clocked out on Off-Production and clocked in on Set-Up or 
Production. The Off-Production Ticket will be filed behind the 
other tickets and may be used for other interruptions if they occur 
on the same lot operation. 

During down-time, if the operator is assigned to another job, 
the Off-Production Ticket will be clocked out, but will remain 
in the visible position in the rack. 
10.0 End of Shift 

Operators will count and report pieces completed to the time- 
keeper. All tickets will be removed from the board and clocked 
out. For each incomplete set-up or production run, prepare a new 
ticket and place it in the rack for use on the next shift. 

10.1 Computing Hours and Pieces 

a. Subtract clocked-in from clocked-out times on each card 
and enter difference in hours to nearest tenth in columns 
provided. 

b. Add each column. 

c. Enter total of both columns in Total Actual Hours. 

d. On Production Cards, extend Good Parts Produced and 
Standard Hours per Hundred and enter Total Standard 
Hours. 

10.2 Up-Dating Manufacturing Release 

Compare the Manufacturing Releases with the Production 
Ticket which show operations completed. Post the date 
completed, good pieces and scrap. If this is the last operation 
on this release, forward the copy by way of the Dispatcher to 
Production Control files. 

10.3 Reconciliation with Payroll Cards 

Compare daily payroll card and job card time totals and 
reconcile. Report major discrepancies to the foreman. For- 
ward Job Cards to Tabulating at end of shift, for key punch- 
ing and preparation of reports. 

11.0 Exceptions and Special Cases 

11.1 Split Lots 

Production Control will occasionally authorize split lots. In 
these cases the letter for the split (A, B, C) will appear in the 
log of the Manufacturing Release. Timekeepers must add 
this split letter below the lot number on all cards affected. 

11.2 Special Order Numbers 

When a change in processing or routing results in assignment 
of a special order number to a lot in production, the foreman 
will inform the timekeeper. He will enter this number in 
the Job Order box on each card affected. This number should 
also be entered in the log of the Manufacturing Release. 



Beaver Alliance Aircraft Corporation 237 

QUESTIONS 

1. What are the systems aspects of this case? 

2. What is the basic problem? 

3. What should Beaver Alliance do to deal with its problem of 
direct labor costing? 

4. How do you evaluate the proposed feeder shop timekeeping 
procedure? In what ways is it superior to the existing procedure? 






17 



A. B. Fleet and Company 



Bonopolis, an eastern city, was forced by expanding population 
and industrial growth, to re-examine their rapid transit system. 
Population had grown from 800,000 to 1,690,000 in ten years. 
Located in the industrial bowl of the United States, this city had 
over 3,000 factories and was located on one of the Great Lakes. It 
was developing as a terminal point for shipping by boat, rail, and 
air. Growth had created a strain on existing transit facilities and 
serious transit problems existed for the bulk of the population who 
were wage earners. 

The Traffic Commission, Planning Commission, and Public 
Utilities Commission shared responsibility for future planning of 
highways and roads and the privately owned transit facilities. Equip- 
ment was old and demands for better service frequently created bitter 
debates in the City Council. 

Typical of their problems were the new developments in outlying 
areas which demanded service. However, the lack of equipment 
and of a plan for operating a system in sparsely populated areas, 
forced transit owners to procrastinate. This created more bitterness 
and public debate. 

In 1954, the matter of providing cheap, effective, mass rapid 
transit became the critical political issue in the city's mayoral 
elections. Since it was no longer possible to delay the adoption of 
a plan, the incumbent mayor and the City Council determined on 
a course of action consisting of the following steps: 

(1) Increase bonded indebtedness of the city by $20,000,000 to 
purchase privately owned facilities and provide modern equip- 
ment and street facilities for mass rapid transit. 

238 



A. B. Fleet and Company 239 

(2) Create feeder systems into new areas to increase the effective- 
ness of rapid transit coverage in the metropolitan area. 

(3) Create a commission of citizens, businessmen, and council 
representatives to study local transit problems. 

(4) Demand a program of the Traffic Commission, Planning Com- 
mission, and Public Utilities Commission which would guar- 
antee continued study of the metropolitan transit issue. 

(5) Merge and extend existing service where possible to reduce 
cost of operation. 

(6) Study fare structures to distribute, fairly and equitably, the 
costs of city-wide operation of the rapid transit system. 

The opposing candidate attacked this plan vigorously. His point 
of view, based upon the activities of other major cities and advice 
from the local experts in this field, indicated some problems in the 
incumbent's point of view. A memorandum prepared by A. B. Fleet 
and Company, experts in the design of mass rapid transit systems, 
is shown as Exhibit One. 

EXHIBIT ONE 
MEMORANDUM FROM! A. B. FLEET AND COMPANY 

Introduction 

There are a number of issues which must be met to begin the develop- 
ment of mass rapid transit for this city. This memorandum describes 
how A. B. Fleet and Company could be of assistance in this program. 

One of the prominent issues requiring early decision is how best to 
combine the major surface lines now in operation. This is today's task. 
But tomorrow's decisions are infinitely more complex. How is this city 
to prepare for decisions that will affect the community ten years hence? 
The design of a mass rapid transit system will be influenced by popu- 
lation growth, industrial development, and changes in land use. There 
is a need for a sound method to assist in the selection of both today's 
and tomorrow's rapid transit routes. 

A. B. Fleet proposes to attack these problems using tools of science 
and electronics. Problems of this type are so complex they defy solution 
by any other means. An example will assist in demonstrating how 
science and electronics can be of help. 

Statement of Problem One 

We would eliminate uncertainty in selecting rapid transit routes, if 
we could see into the future. A glimpse into the future would tell us 



240 A. B. Fleet and Company 

where communities would spring up or where industrial development 
might thrive. So let us postulate a power that can make changes in 
population, land use, and economic growth. Our large area now be- 
comes a constantly changing panorama. People, their homes, their 
places of work, and industrial firms respond to new conditions. The 
face of our city in 1954 and 1980 appear considerably different. 

If we had the power to see our city as it will be in 1980, we would in- 
corporate many changes in our existing plan. The more visibility we 
can obtain in 1954, the more certain we can be that we have provided 
for the future, and minimized the possibility of mistakes. The question 
is, Are there in existence technical tools that will assist us in solving the 
route selection problem? 

The answer is yes! 

The Use of Computers and Simulation 

A. B. Fleet has actually solved similar problems which required the 
ability to predict far in the future. This has been done using a tech- 
nique known as simulation. Simulation is done on a computer which 
has the ability to process data at lightning speeds. Because of these 
electronic speeds, it is possible to compress many years of experience in 
a few seconds of time. Simulation utilizes the high speed of the com- 
puter to reproduce in a small fraction of time the experiences of many 
years. In a recent simulation problem, A. B. Fleet squeezed two hun- 
dred years of inventory experience into less than two hours! 

Simulation can assist in determining the capacities of rapid transit 
routes under different conditions. This tool has already been used to 
assist in solving traffic and transportation problems. In a freeway prob- 
lem, traffic was represented by electrical impulses moving at very high 
speeds within the computer. Given the distances between entrances and 
exits, a variety of speeds for each car on the freeway, and various load 
conditions depending on the time of day, it was possible to reproduce 
the traffic flow the freeway was able to carry, before the freeway was 
built! 

Simulation can be used to study variables that effect the design of 
rapid transit systems. In the process of doing this, the important vari- 
ables are isolated and their effect on the system reproduced. Therefore. 
the first step is to identify as many variables as possible. Statistical tests 
must then be made to determine which variables are important and 
which are unimportant. Because we are using a computer, we can deal 
with complex, real-life situations containing many variables. Problems 
with a large number of variables which require an analytical solution, 
might require years — perhaps are impossible — to solve without a com- 
puter. 



A. B. Fleet and Company 241 

Simulation allows us to deal with minute pieces of data. Without the 
computer, it might be necessary merely to approximate the solution to 
a problem, because the analyst would be unable to cope with a large 
volume of data. The solution would have to be constructed from gross 
measures as opposed to fine measures. With the help of a computer, it 
is possible to solve origin and destination problems using very small 
zonal areas. By analyzing data in detail, the possibility of errors in route 
selection is considerably reduced. 

Simulation allows us to study rapid transit in competition with other 
modes of travel. The number of rapid transit commuters will be de- 
termined by many variables, such as cost, speed, accessibility, and con- 
venience. Given a set of facts on a proposed route, a computer can 
reproduce the conditions under which any segment of the system would 
operate. In addition, it could also simulate the competitive modes of 
travel. This would be invaluable in solving problems, such as how to 
attract more users to mass rapid transit. 

Simulation would make it possible to consider many alternate routes 
and rapid transit systems. Given a set of data representing a tentative 
route, the computer can examine several alternate means of traveling 
from an origin to a destination. In this way, many alternate rapid 
transit systems could be tried and the best routes selected. The ability 
to examine many rapid transit routes at frequent intervals makes it 
possible to do better transitional planning. 

Summary 

Simulation is a scientific tool which can be put to use in the solution 
of rapid transit problems. A. B. Fleet has made successful use of simula- 
tion as a tool in the solution of nonscientific problems. Simulation on a 
high speed computer can determine route capacities under different 
conditions of load. The many variables affecting the design of rapid 
transit systems can best be studied using computer simulation. The 
high speed of simulation makes it possible to deal with large quanti- 
ties of data. The competitive problems of rapid transit versus other 
modes of travel lend themselves to analysis through the simulation 
technique. Simulation makes it possible to select at frequent intervals 
the best route or best system out of many alternate routes or systems, 
and thus provide adequate transitional planning. 

Statement of Problem Two 

The design of a rapid transit system for this metropolitan area re- 
quires a variety of special skills. There are many technical problems 
associated with each phase of the program which must be solved to 



242 A. B. Fleet and Company 

bring about a successful system. Not only must these problems be antici- 
pated, but it is essential that the entire program be expertly guided 
through the many difficult decisions that will require action. 

A wide range of costly decisions will be required to activate an up to 
date rapid transit system. There will be a requirement for data support- 
ing these decisions that will minimize the possibility of errors in judg- 
ment. There is an associated problem of determining what part of the 
available data is valid, and what additional data are required to com- 
plete the evidence. 

The design of pilot systems and the selection and development of 
data and technical advice are required to coordinate and integrate 
this complex program. This is a method of operation and project 
organization familiar to A. B. Fleet and Co. As technical advisor to 
the many industries, A. B. Fleet coordinates and oversees the activities 
of many large corporations. Our efforts have been directed at solving 
some of the most complex scientific and engineering problems ever en- 
countered. In order to outline how such a technical advisor might oper- 
ate for the City's Transit Authority, a tentative plan of action can be 
suggested. 

Plan of Action for the Technical Advisor 
Phase One: Organization 

Phase One is a preliminary study which has the goal of assembling 
available information on the local rapid transit problem. 

The unique aspects of the city's problem would be isolated and ex- 
amined. The many agencies that have an interest in advancing this 
program would be consulted, and liaison established on a regular basis. 
The data available on studies conducted in other cities would be as- 
sembled and sifted to cull information that is of local interest. The 
priority of effort would be established. 

A method of operation and liaison would be established with the 
Transit Authority, so organization of this project would be responsive 
to their needs. 

Phase Two: Preliminary Systems Design 

Phase Two is the systematic study of data relating to the problems 
of merging existing rapid transit lines and preparing for future systems. 

Men experienced in traffic and transportation problems would study 
available data to develop the system limits. They would develop criteria 
which would hold for short range and long range planning. They 
would develop the initial concepts to assist in the design of the rapid 
transit plan, and would test these concepts by the proper statistical tech- 



A. B. Fleet and Company 243 

niques. They would draw a plan for conducting the scientific portion 
of this study so their work would be finished when needed. 

Parallel with this effort, other personnel would begin the collection 
and integration of field data. The following are typical questions which 
would be answered in this phase: 

Land use and population: How does land use affect the design of the 
system? Where are populations currently concentrated? Where are popu- 
lations moving? What residential areas are becoming commercial or 
industrial? How are land use, population, and circulation interdepend- 
ent? 

Origin and destination: Where do people live and work? How do 
they travel now? What will be the influence of a completed freeway or 
rapid transit system? What are the other "competitive" modes of travel 
and how do they affect the systems design? 

Route sections: What should the dimensions of key thoroughfares be? 
What are the possible rapid transit routes? What plans exist for wide- 
ning or otherwise altering present street systems and how will they affect 
systems design? 

Measures of actual flow: What are the capacities of present key thor- 
oughfares? How will these change as population or land use shifts? 
What are desirable routes based on actual flow and how well do they 
correlate with other data? What can be predicted about future route 
locations based on anticipated changes in flow? 

Rapid transit facilities: What are the present and planned facilities? 
What new facilities are required to fill the needs of the community? 
What are the present and future economic requirements of the system? 
What are the alternate modes of mass rapid transit and the costs inherent 
in each? Who are the proposed contractors for new transit systems and 
what are their recommendations? 

Geographical and geological factors: What are the physical charac- 
teristics of the area and how do they dictate the type of rapid transit 
systems? What are the possible modes of mass rapid transit? 

Sociological and psychological factors: How can we be certain new 
facilities will be used as planned? Will new transit systems be accepted 
by the population? Why are some routes or facilities better used than 
others? How can the economic success of a proposed system be tested 
prior to installation? 

Simultaneous with the collection of field data and the initial scientific 
studies, a financial and legal survey woidd be conducted. Real estate 
data would be collected and studied to determine its influence on the 
design of a new system. Federal, state, county, and municipal legisla- 
tive requirements would be studied to assure adherence to existing 
codes. We would develop financial data on the cost of alternate systems 
and accumulate facts to support proposed revenues for these systems. 



244 A. B. Fleet and Company 

Phase Three: Analysis and Final Systems Design 

Phase three begins when all of the necessary data could be assembled 
and analyzed as a unit. 

At this time, the assumptions of the proposed systems would be 
tested and changes made in the tentative system design. Errors would 
be determined and corrections built into the new design. Simulation 
would test and improve the system design. Development of current 
systems and anticipated extensions or revisions to proposed systems 
would be tested. 

Technical advisers would recommend the best system for each selec- 
ted time period in this phase. Actual routes would be identified and a 
total area plan, incorporating the necessary rapid transit requirements, 
would be submitted. 

Summary 

Design of a mass rapid transit system requires a variety of technical 
skills. Before costly decisions are made, large quantities of data must 
be accumulated and analyzed. A. B. Fleet and Company has successful 
experience in the coordination and integration of complex programs 
such as this, and has developed a method of operation specifically de- 
signed to support systems installations. As technical adviser, A. B. Fleet 
would prepare a plan of action geared to the requirements of this 
city's Transit Authority and the needs of this city. This activity would 
progress through three phases: organization, preliminary systems de- 
sign, and analysis and final systems design. The end result would be the 
determination of area mass rapid transit requirements, based upon 
current factors, and projected into the future, based upon scientifically 
analyzed trends. 

Conclusion 

The problems of unifying present surface systems and determining 
future mass rapid transit requirements require a set of special talents. 
A. B. Fleet and Company would attack these issues on two fronts: 

1. A. B. Fleet proposes to add the elements of science and elec- 
tronics to assist in solving complex problems, such as the selec- 
tion of rapid transit routes. 

2. A. B. Fleet proposes to become the technical adviser to our 
Transit Authority because of its qualifications and recognition 
as a leader in the field of systems coordination and integration. 



A. B. Fleet and Company 245 

QUESTIONS 

1. What is the system under study? What are the major subsystems? 

2. What are the objectives or outputs at the various system and 
subsystem levels? 

3. Who are the systems purchasers at the various system and sub- 
system levels? 

4. What kinds of criteria might be developed to measure the effec- 
tiveness of one rapid transit system over another? 

5. Compare the two programs. How does the Fleet memorandom 
stand up as a communication tool? 

6. What is the problem? 



18 



Wesley Engineering, Inc 



Wesley Engineering 1 was a multi-division organization performing 
contract engineering work for a number of industries. Divisions 
were specialized for work in the fields of aircraft, shipbuilding 
civil engineering, plant construction, and atomic energy. Each 
division was autonomous and was staffed to operate as an independ- 
ent entity. Administration, sales, and technical work were tailored 
to the requirements of each division. Contacts between the oper- 
ating staffs of divisions were rare except when large projects de- 
manded the formation of teams to work on the same contracts. At 
the time of this case, in 1958, each division employed about 200 
engineers and technicians. In each division, about 50 administra- 
tive, non-technical personnel supplied the services which were re- 
quired. Total dollar volume was $21,000,000 per year. 

Like most firms in this field, Wesley saw its most spectacular 
growth starting in 1940. As the need to meet more complex and 
profitable opportunities arose, teams of specialists were formed 
from within and outside the organization. With each new contract 
in some specialized field, the size of a team would grow and its 
interests become more foreign to existing divisions. The result of 
this was the creation of new divisions and the development of sup- 
porting services which invariably paralleled the other divisions. 

Fredrick Wesley advocated divisional autonomy because he 
recognized the need to nurture the individual interests and ambi- 
tions of his division managers and their staffs. Moreover, he found 
that the conflicts which customarily arose in closely held, non- 
autonomous divisions, were absent under his organization. Since 

i See Chapter 5, page 81. 

24 6 



Wesley Engineering, Inc. 247 

each division was on an individual profit and loss basis, its indepen- 
dence assured the minimum of shared costs, except for the charges 
for corporate staff management. The Wesley organization chart 
(Figure 18-1) is shown as Exhibit One. 

EXHIBIT ONE 



WESLEY ENGINEERING SERVICE, 
ORGANIZATION CHART 



INC 



CORPORATE 
CONTROLLER 



DIVISION 
MANAGER 



PLANT 
CONSTRUCTION 



F. WESLEY 
PRESIDENT 



CORPORATE 

CONTRACT 

ADMINISTRATOR 



DIVISION 
MANAGER 



CIVIL 
ENGINEERING 



CORPORATE 

SALES 

DIRECTOR 



CORPORATE 

ENGINEERING 

DIRECTOR 



DIVISION 
MANAGER 



AIRCRAFT 
ENGINEERING 













DIVISION 
MANAGER 


ATOMIC 
ENERGY 






I 




1 


ADM. 




TECH. 














SALES 





DIVISION 

MANAGER 


SHIPBUILDING 
CONSTRUCTION 






r~ 


1 


ADM. | 


1 TECH. 












SALES 





Fig. 18-1. 

The corporate structure was maintained to provide control over 
policy matters. Once monthly, the division managers met with 
Wesley and his four man staff as the corporate steering committee. 
Likewise, as shown in Exhibit Two (Figure 18-2), three of the four 

EXHIBIT TWO 

WESLEY ENGINEERING SERVICE, INC. 
CORPORATE CONTROLS ORGANIZATION 



CONTRACT 
ADMINISTRATION 



PLANT 
CONSTRUCTION 



CIVIL 
ENGINEERING 



AIRCRAFT 
ENGINEERING 



CONTROLLER 

STAFF 
SERVICES 



ALL 
DIVISIONS 



ENGINEERING 

STAFF 

SERVICES 



PLANT 
CONSTRUCTION 



SALES 

STAFF 

SERVICES 



ALL 
DIVISIONS 



Fig. 18-2. 



248 Wesley Engineering, Inc. 

corporate staff officers met with the divisional executives whose 
activities were under their nominal control. Four divisions were 
free of even this control in the engineering area, because of highly 
specialized problems beyond the corporate engineering staff's abil- 
ity to give effective supervision. The shipbuilding and atomic 
energy divisions were also independent of Corporate Contract Ad- 
ministration because their work was entirely in the research and 
development area. These departures from the format of control, as 
it was originally conceived, were difficult to obtain and were under 
constant surveillance. 

The methods of keeping costs and handling Contract Administra- 
tion were not identical in each division. The need to meet circum- 
stances unique to each division had created the opportunity for each 
division to solve its problems in its own way. Thus, there were five 
cost systems, five contract administration systems, and so on, each 
of which were supervised at the corporate level to conform to some 
generalized model acceptable to top management. 

Typical of this situation was the forecasting process. Each Divi- 
sion had a unique method of doing its advance planning. Semi-annual 
reports were prepared by the administrative directors of each di- 
vision indicating the forecasted manpower requirements by proj- 
ect and by each six month period into the future for two years. 
Using standard rates of pay for the various job classifications and 
calculating the productive work hours in each period, they calculated 
the direct labor hours. Overhead accounts were projected, based 
upon history and anticipated changes in cost. From these figures, 
overhead rates were calculated to provide a method of quoting on 
new contracts. From this same information, sales, cost of sales, and 
profit were projected. 

There were many difficult problems associated with these calcu- 
lations. Productive hours depended upon several variables, such as 
vacations, holidays, anticipated overtime, sick leave, and transfers 
in and out of a division. To bring net total hours down to a con- 
servative estimate of productive hours, the employee structure on 
a department and section basis had to be analyzed. This proved to 
be an exhaustive process which was difficult to update. 

The analysis of average wage rates was time consuming. The 
process of obtaining raw data on the number of employees in each 



Wesley Engineering, Inc. 249 

job classification came from section heads and received the approval 
of department heads. But each set of wage rates required modifica- 
tion for anticipated rate changes over this extended period. Further- 
more, frequent changes in contract requirements would force sec- 
tions to be reconstituted on a different personnel base. As a result, 
very gross measures were used in the determinations of average 
wage rates, and reports were never an accurate reflection of the 
status quo — or of the future. 

Forecasts were done in each division on a manual basis. About 
three months were required to prepare a forecast from new head- 
counts. Material from each division required extensive revision 
before the preparation of a corporate forecast was possible, since 
information was not parallel. Among many problems were those 
occasioned by layoffs in one division while costly hiring was taking 
place in another. 

In June of 1957, at the suggestion of the corporate controller, 
Wesley decided the time had come to tighten up the forecasting 
area of control because of changes that were taking place in many 
industries. He put the problem up to his division managers at their 
regular meeting, stating four principle needs: 

1. Reduce personnel pools, particularly administrative. 

2. Reconstitute technical sections on minimum levels. 

3. Develop a reporting system sensitive to quarterly changes. 

4. Simplify method of forecasting, so interim forecasts are possible. 

Division managers conferred with their administrative staffs im- 
mediately, in anticipation of another meeting to discuss what steps 
should be taken to implement Wesley's suggestions. When the 
second meeting was called, the division managers presented an ad- 
ditional series of problems for discussion: 

1. Division autonomy was being invaded. 

2. Corporate forecasts were unnecessary. 

3. Divisional problems precluded a uniform forecasting system. 

4. There were no known methods of mechanizing forecasting. 

5. Accuracy and reliability of forecasts could be questioned, but 
whether they were inaccurate or unreliable to any significant 
amount was unknown. 



250 Wesley Engineering, Inc. 

6. The value of forecasts was considerably diminished unless they 
were used in a gross way. 

Wesley agreed to make no decision regarding how each division 
must forecast, but obtained agreement that, even if the methods 
were not uniform, a better model was essential. Accordingly, he 
asked the corporate controller to monitor a study to be undertaken 
by his staff on the problem of forecasting. He was asked to start at 
the earliest moment and to advise Wesley as soon as he had a plan 
to present to the division managers. 

About four months after this date, the division managers and the 
corporate staff members heard a talk given by John Wills, the proj- 
ect leader on the forecasting study. The talk is reproduced as Exhibit 
Three. 



EXHIBIT THREE 

PROPOSED FORECASTING SYSTEM 

A TALK BY JOHN WILLS 

The study of forecasting at Wesley began with the examination of 
the questions, "Why is it necessary to forecast? What is a forecast?" I will 
answer these questions in terms of the functions of forecasting. 

A basic function is to project sales and expenses so the flow of funds 
and financial requirements can be anticipated. Equally important, the 
projection of sales and expenses previews the expected profit. Contract 
Administration requires forecast information to prepare proposals and 
bids. Budgeting and future planning are made simpler and more effec- 
tive if Management is working from adequate information. Facilities 
and equipment planning will be enhanced if forecasting procedures 
provide reliable information when it is needed. From this we can con- 
clude a working definition of forecasting: Forecasting is the process of 
assembling current information and relating it to the future needs and 
requirements of a company. 

As a result of investigating these basic questions, we learned the 
symptoms of our inadequate system. Prominent among these symptoms 
was the problem of how long it takes to produce a complete forecast. It 
is obvious that the rapid growth of this company has made the produc- 
tion of forecast information more difficult and time consuming. This 
problem in its most acute stage is typified by recent forecasts which 
were invalid by the time they were completed — because personnel mix 
and sales requirements had changed markedly while the forecasting 
computations were in process. A second symptom was obvious because 






Wesley Engineering, Inc. 251 

the week-to-week changes in sales requirements could not be success- 
fully incorporated in the forecasts without indefinitely delaying their 
completion. It is therefore necessary to take the position that the 
present system is incapable of producing current, valid information. 

A number of symptoms can be grouped under the current method of 
producing the forecast information. First among these, the corporate 
controller finds it increasingly difficult to maintain the total work load 
due to the large volume of man-days consumed in forecasting. We have 
a completely manual system which requires large volumes of simple 
arithmetic computations. An enormous variety of source documents 
are required to produce the forecast. The necessity of maintaining a 
high level of accuracy dictates an excessive amount of double checking, 
verifying, recapping, and rearranging. 

The time required to make a complete forecast can be laid out in 
seven steps: Department headcount and direct costs require five days. 
Division wage dollars and overhead expenses require ten days. Cor- 
porate allocated overhead expenses require fifteen days. The compu- 
tation of division cost of sales and overhead rates take an additional 
five days. Corporate general and administrative expenses require fifteen 
days. To digest this load of data and obtain division sales and profits 
takes five days To produce the final document, the corporate forecast, 
requires an additional five days. It therefore appears that forecasts 
cannot be produced in less than sixty working days or twelve weeks. 
Clearly what is needed is a simple, more rapid way of doing this job! 

The first look at the current forecasting procedures was aimed at 
identifying the problem area. To do this, we began a step-by-step process 
of assembling the information that would describe the present system. 
If we had correctly assessed the symptoms, it was now time to explore 
the causes and get to the root of the problem. We asked, "What is it we 
are trying to forecast?" It is clear we are trying to build a forecast from 
the most elemental data. We want to know how many personnel are 
required for each project that is either in-work or anticipated, so we can 
allocate a specific personnel mix to each forecasting period in the future. 
In addition to predicting the headcount, the present system is allocat- 
ing divisional overhead and corporate overhead to each forecasting 
period, so the overhead rates can be used to determine cost of sales and 
profits. 

If there is any single factor that makes the forecasting problem diffi- 
cult, it is the problem of how to obtain the proper average wage rate by 
job classification for each project in each forecasting period. These 
figures are critical in the forecasting procedure because they are used in 
the later computation of overhead rates. Let me lead you through the 
present method of computing average wage rates and overhead rates. 

What is immediately clear is that there are four factors that affect 



252 Wesley Engineering, Inc. 

each and every project for which manpower is to be allocated. Present 
wage rates must be modified by anticipated merit increases. Present 
headcount by job classification must be amended by the anticipated 
headcount in each job classification. When all of the data are accumu- 
lated, it is possible to assess for each project in each forecasting period 
the average wage rates by job classification. Note, however, that each 
project requires a wealth of data, and each project, has to be added into 
the total picture, a piece at a time, on a completely manual basis. It is 
no wonder that it requires three months to complete a forecast, and 
that errors creep into the maze of arithmetic calculations. But now we 
can proceed to the next example: Given the average wage rates, how is 
the overhead rate computed? 

You will remember that to make a successful forecast we allocate sales 
and expenses to each project in each forecasting period. This means 
that the various overhead items, such as indirect labor, non-labor ex- 
penses, and payroll expenses must also be accumulated in a similar 
fashion. In fact, this is being done on the same laborious manual basis 
we observed earlier. The corporate controller allocates dollars to each 
project in each forecasting period to spread properly the overhead costs 
that are anticipated in the future forecasting periods. However, before 
overhead rates can be obtained, the average wage rates must be multi- 
plied by the anticipated direct headcount, project by project. This 
produces the total direct labor wage dollars for each project in each 
forecasting period. 

With all the pertinent overhead costs assembled for each forecast 
period, the final computation is to divide total overhead dollars by the 
total direct labor wage dollars for the identical periods. The resulting 
overhead rates are now ready for use in allocating overhead dollars to 
each project in each forecasting period, with the intent of determining 
the cost of sales and potential profit. This is not the sum total of what 
is required in a new system. There are other inputs, such as productive 
work hours which require extensive computation, which we will not 
explore this afternoon. But, in summary, we can say that the system that 
exists today has been completely outgrown by Wesley, and that it is 
altogether inadequate for the present and future purposes of Manage- 
ment. What is the next step? 

A computer application is indicated by the volume of data that re- 
quires processing. Current requirements are set at 7,500 input entries 
and over 40,000 output entries. It is estimated that by 1960, there will 
be over 50,000 input entries, and over 250,000 output entries, if Manage- 
ment does not require additional information or increased frequency of 
reporting. This conclusion is reinforced by the need for faster reporting 
without increasing, if possible, the cost per unit of operation. If in- 
formation could be made available faster, it is obvious we would obtain 



Wesley Engineering, Inc. 253 

many of the advantages as they were outlined at the outset of this report: 
current and reliable information on cost of sales and profits to guide 
Management in its decision making. However, in making a computer 
application, we have several alternatives. The first of these is to make 
a one-for-one changeover and to go through each of the processing steps 
as they are being done — but to do them faster. This does not seem 
practical in view of the large volume of data to be processed and does 
not satisfy the criteria of a more simple system. 

We were fortunate to obtain the services of one of our technical staff 
experienced in the use of management statistical tools and techniques. 
Working closely with him, we reviewed the present state of the art of 
forecasting and pointed out the weaknesses that existed. A detailed 
examination of the data was begun, and soon it appeared that there was 
a novel approach to the problem of simplifying the computational tasks. 
Trial and error eventually revealed that the statistical technique of 
regression analysis could be applied successfully to this problem, with 
improved results. The techniques of regression analysis reveal that, as 
headcount increases, there is a fixed relationship with total wage dollars 
per hour. Points can be plotted on a chart to represent the relationship 
between headcount and total wage dollars per hour in different months. 
The line drawn between the points on the chart is represented by an 
equation. The value of being able to write an equation representing this 
regression line is, of course, convenience. With the equation, a person 
can quickly do the arithmetic computation to obtain total wage dollars 
per hour, without referring to the chart at all. This equation, 

$85 + $2,623 (HC) = Total wage dollars per hour, 
where, HC = Headcount, 

appears to satisfy our system requirements. It was now necessary to test 
this equation to determine its reliability. To do this we checked the 
coefficient of correlation, which proved to be .95. You will be interested 
to hear what the regression formula revealed when it was compared to 
the actual costs of the periods tested. We tested six past monthly periods. 
For each of these periods we moved from the headcount to the total 
wage dollars which we would have forecasted by our regression line. 
This deviation was found to be well within the requirements of the 
system for accuracy. 

It is proper for us to ask the question, "Exactly what advantage does 
the regression analysis technique give the proposed forecasting system?" 
It is clear that the regression technique simplifies the computation of 
total wage dollars per hour. What formerly had to be computed by 
lengthy, tedious, manual means, requiring many inputs, is now pro- 
duceable with only two inputs. The elimination of a cumbersome 
technique has, in addition, created the possibility that the new system 



254 Wesley Engineering, Inc. 

might operate more economically. Accordingly, this technique will be 
used elsewhere in the system to determine overhead dollars. The other 
improvement is the increased speed with which computations can be 
made. Even though the new system is still on a manual basis, it is 
estimated that the reduction in time is 30 per cent. What the computer 
might add in terms of additional speed remains to be seen. 

You have heard the important details of the proposed forecasting 
system. Now we can look at the "big picture" of the new system and see 
how it can function on a more simplified basis. You will observe that 
we now need only two basic inputs — headcount and total available 
hours. In the old system, we needed not only these two inputs but also 
a considerable number of other basic inputs. From only headcount and 
available hours, we are now able to arrive quickly at direct and indirect 
wage dollars and all the overhead expense dollars. Once this informa- 
tion has been obtained for each project in each forecasting period, it is 
a simple matter to compute future overhead rates, cost of sales, total 
cost of sales, and sales dollars. 

To determine the reliability of the systems design, the proposed sys- 
tem has been operated manually. In this way, each step of the data proc- 
essing was checked to be certain of its effectiveness in the preparation 
of the forecast. If this system is acceptable, the final phase of our stud\ 
will be divided into four activities. The first activity will be the writing 
of a program in anticipation of leasing time on a computer. The second 
activity will be concerned with the design of new input documents and 
output reports to accommodate a more simplified presentation of fore- 
casting results. The third activity will surround the preparation of 
written procedures to describe adequately all phases of the new fore- 
casting procedure. The fourth activity will be centered about the de- 
velopment of computer routines for controlling and checking the equa- 
tions used in the regression technique. In addition, we will undertake 
the development of routines so the computer can develop the regression 
formulas without human intervention. 

We were given a general problem. The problem was to reduce the 
clerical and administrative man hours required to prepare a forecast. 
As a result of our study and planning, the forecast for forecast prepara- 
tion time may be reduced from 12 weeks to 2 weeks. In effect, the 
computational bottleneck will be broken and the goal of the study will 
be reached by the use of a computer. However, there will be some addi- 
tional benefits that can be classed as unanticipated by-products. A sav- 
ings of $80,000 is estimated over the next four year period. Reliable, 
accurate, and current information will be available. The new system is 
flexible enough to expand and contract with changing business needs 
and conditions. These studies have opened up new areas where the 
regression technique can be used. Additional data can now be obtained 



Wesley Engineering, Inc. 255 

at a very small marginal expense. We now have a simple and quick 
method for obtaining revised forecasts. Finally, proposals can now be 
prepared on a more competitive basis. 



QUESTIONS 

1. How would the systems analyst state the problem of Wesley 
Engineering, Inc.? 

2. How well does the proposed forecasting system fill the needs of 
the corporation? 

3. What are the systems aspects of this case? 



; 



19 



Davis Engineering 
Company 



C. R. Davis started in the business of making tools and dies in 1907. 
With Henry Russell, he gradually built a small, successful company., 
over a period of years. In 1927 Russell died, and Davis bought out 
his interest. The business continued to grow. 

Davis Engineering Company built tooling from small die sets to 
large press dies to accommodate the largest hydraulic presses. Davis 
put his efforts into the operation of the shop and the supervision of 
the engineering efforts. His pride was in the reputation for skill and 
quality for which his firm was known. The emphasis was not heavy 
in administration and organization. Older employees voiced the 
well-accepted attitude that the firm's steadily rising dollar volume 
and profit were reflections of confidence of old customers. 

In 1932, Davis became interested in the potential offered by the 
air-frame industry. He saw not only large volumes of tools of all 
kinds, but vast possibilities in jigs and fixtures. Always on the alert 
for a new technical challenge, Davis was able to obtain contracts 
because of his excellent reputation. The first few contracts were 
very costly and heavy losses pulled him farther and farther into the 
shop operation. 

Customer contact was shifted from Davis to experienced office 
and engineering personnel who took on the job of liaison between 
the Davis organization and the customer's engineering department. 
Gradually, these part-time liaison men had to be replaced in the 
organization, because they found more and more to do to keep the 
new customers well satisfied. When this shift in personnel took 
effect in 1936, the two men, Joe Hardy and Andy White, were given 
the title of sales manager. They were given separate offices, secretar- 

256 



Davis Engineering Company 251 

ial help, and their duties were officially recognized. The sales or- 
ganization had officially come to life. 

By 1940, Davis Engineering Company was no longer the same, 
small business of the early thirties. They had moved into a 100,000 
square foot building and their backlog with large manufacturing 
companies was almost $10,000,000. Hardy and White were regarded 
as the high powered salesmen who had "made" Davis Engineering, 
although Davis himself was the major factor in decision making 
and the final authority. There was no disputing with Davis why it 
was possible to get repeat orders. As a consequence, the sales or- 
ganization did not grow, and the place of sales in the organization 
was never well defined. 

When World War II ended, Davis was operating an organization 
of twelve hundred, and was considered among the most successful 
in the country. Prior to the Korean War, there was a severe drop in 
business. But because of certain manufacturing contracts which 
Davis obtained through his connections, the recession was short 
lived. With the resumption of defense activities, Davis again built 
its volume substantially. In the meantime, rising overhead had in- 
creased their break-even point. The manufacturing business was con- 
tinued on a small scale. 

In 1955, Hardy was forced to resign because of poor health. White 
continued in his former role without assistance, since the backlog 
was at an all-time high of $22,000,000. This was regarded as a two 
year load, with no apparent letup of business in sight. 

In the summer of 1956, abrupt cancellation notices rocked the 
Davis organization. Overnight, nine major projects were halted by 
prime contractors who had received cancellations from the govern- 
ment. The backlog dropped to $5,000,000. It was clear that there 
would be additional business, but White did not think the ex- 
panded operation could wait for new contracts to be let. A sub- 
stantial layoff was in prospect. 

Davis was bitter over what he termed "customers that blew hot 
and cold." He determined to do something that would insure busi- 
ness volume and insulate his operation from being completely 
devoted to industries that were dependent on the armed services. 
He studied the situation briefly, and announced a new policy where 
military business would be only 60 per cent of the total sales volume. 



258 Davis Engineering Company 

As a first step to diversify the range of possible customers, Davis 
directed White to develop a "Road Show." He explained his idea 
in this way: 

What we need are more customers! What we want are less military customers! 
The problem is really not very complicated. We have to let the commercial 
customers know we're alive, and we'll get our share of whatever business we 
can handle. Now, I think the military contracts we've been working on are 
a fine recommendation. If we can get this message across to commercial cus- 
tomers, we're back in business in a big way! What could be a better way of 
doing this than to write letters to the biggest manufacturers all over the 
country, and tell them we'll bring our bag of tricks to their own back door 
for them to see? I've told Andy White to organize a Road Show for us to look 
at, and we'll start the performance rolling in thirty days. 

White began the process of assembling a Road Show. However, 
shortly after he began work on this project, it appeared to raise 
many questions. The further he went, the more it became obvious 
that the Road Show was one facet of a many-sided issue. He finally 
decided that before he went any further, he would have to call a 
halt and report his findings, despite the fact that they would be 
greeted with some scepticism. This report is attached as Exhibit 
One. 

White was surprised when Davis reacted to this report by com- 
missioning a consulting firm to look into the sales problem. The 
consultants made a survey and submitted a report which has been 
abstracted as Exhibit Two. 



EXHIBIT ONE 
OUTLINE OF INVESTIGATION BY ANDY WHITE 

I. What are the essentials of a Road Show? 
II. Questions posed by the Road Show 

III. Tentative conclusions regarding Road Show problems 

IV. General conclusions and definition 



I. What are the essentials of a Road Show? 

A. A format or program tailored for each presentation 

1. Scope (many alternates depending on audience and location) 
a. A.M. through Luncheon 

(1) Semi-technical talks 

(2) High level luncheon speaker 






Davis Engineering Company 259 

b. A.M. through P.M. 

(1) Semi-technical talks in A.M. 

(2) High level luncheon speaker 

(3) Activity with accent on audience participation in P.M. 

c. Dinner and evening presentation 

(1) High level guest speaker 

(2) Semi-technical talks 

(3) Social period 

2. Content (dependent upon scope) 

a. Written material 

(1) Specially edited documentation of program 

(2) Brochure of activities 

(3) Give-aways 

b. Oral presentations 

(1) Talks 

(2) Discussion groups 

(3) Question and answer sessions 

c. Demonstrations 

B. Support necessary to make a Road Show successful 

1. Visual aids 

a. Films 

b. Slides 

c. Charts 

d. Exhibits 

2. Literature 

a. Non-technical and technical material 

b. Brochures, invitations, ads 

3. Publicity 

a. Before and after Road Shows 

b. Newspapers, magazines, and so on 

c. Direct mail 

4. Promotion 

a. Personal letter campaigns 

b. Personal contact campaigns 

c. Group contact campaigns 

d. Direct mail campaigns 

5. Follow-up 

a. Direct mail 

b. Personal calls 

c. Social or semi-social activities 

II. What are some of the problems posed by the Road Show? 
A. Audience 

1. How can we be sure the audience contains potential cus- 
tomers? 



260 Davis Engineering Company 

2. Where do we put on the show? West Coast? Nationally? 

3. Mixed groups? Special groups? Individual firms? 

4. What impressions do we wish potential customers to have 
after we leave? What levels of management are we trying to 
reach? 

5. Can Road Shows be used as a means of generating good will 
and interest in groups who may not be direct, potential cus- 
tomers? 

B. Company operation 

1. How does the content emphasize our experience and capabil- 
ities? 

2. How do the technical personnel and techniques we employ 
support the Road Show? 

3. What tools do we need? 

4. How do we follow up and close a sale? 

C. Why a Road Show? 

1. What is the fundamental purpose? 

2. What is the budget? 

3. Who will put on the show? 

4. Why do we do this in preference to something else? 

5. What other things might we do? 

6. What is the theme? 

III. What tentative conclusions can we make regarding the Road Show? 
A. Audience 

1. We need the "rifle" approach as opposed to the "shotgun." 

a. Mixed groups have mixed interests 

b. The more specialized the group, the better is our chance 
of creating an impression 

c. The ideal audience is one which is interested and has in- 
vited us in; no door opener required and existence of a 
need already established. 

d. Road Shows are a slow, expensive way to make contacts, 
as opposed to ads, letter campaigns, and other personal 
means. 

e. If the Road Show is put on in response to a request for 
more information, reasonable, tastefully stated, selling 
points will not be objectionable. 

2. Attendance at a Road Show will be poor if it is used as a sales 
pitch. 

a. High level executives will not attend to be "sold." 

b. Attendance will be based upon their need to know. 

c. We must satisfy a real need in our show. 

3. The audience becomes our salesmen if the Road Show is suc- 
cessful. 



Davis Engineering Company 261 

a. The needs of each group or potential customer must be 
analyzed. 

b. A concerted, organized, well-timed plan must be devel- 
oped to turn the potential into a sale. 

c. The Road Show must be a smooth, professional, well- 
organized sales effort calculated to leave a strong impres- 
sion of good-will and confidence in our ability. 

d. Our Road Show contacts will die on the vine if we don't 
continually supply them with ammunition and point 
them more and more toward a specific goal. 

4. Analysis of contacts in 1955 indicates we have a larger audi- 
ence in the East than on the West Coast. 

a. We must create a suitable means for operating in the 
East, to overcome most of the current objections. 

b. If we cannot set up suitable means of operating nation- 
ally, expensive national advertising should not be con- 
sidered. 

c. Pursuit of business in the eleven western states alone, re- 
duces our chances of obtaining our optimum volume by 
66 per cent. 

d. Organization modifications should accompany adoption 
of a sales policy. 

e. Competitive considerations are effected by the selection 
of a policy. 

5. The Road Show idea must be used in as many ways as its flex- 
ibility will permit. 

a. Community or trade groups 

b. Non-competitive professionals and consultants 

c. Interested, but non-client industrial groups 
B. Technical Groups 

1. We can only talk about work we have done or work we are 
prepared to do. 

a. Material must be specially tailored to the potential 
client's or group's needs, from some basic format. 

b. To the extent that we have specialized portfolio, this will 
have a bearing on the type of personnel we hire. 

c. There must be some assurance that the potential customer 
has a real interest in what we have to sell. 

2. Technical groups must support the Road Show. 

a. Members of the staff must be used where necessary to 
make the shows more effective. 

b. Articles, talks, and personal contacts with existing or 
potential clients must consider the necessity of endorsing 
and promoting the Road Show. 



262 Davis Engineering Company 

c. Additional permanent or semi-permanent assignments to 
the sales effort will increase amount of activity. 

d. Every staff member must understand the sales strategy, 
his role in sales, and how to conduct himself in the sales 
situation. 

3. How will plans for a Road Show be laid? 

a. Principal areas of customer interest must be built into 
basic formats of talks, written material, and exhibits. 

b. Some appreciation of actual customer-Davis relationships 
must be built into the Road Show by personnel preparing 
the program. 

c. Special visual aids will supplement standard types of 
sales tools which explain our underlying philosophy and 
general services. 

d. A sales strategy must be conceived for each potential 
customer in which the Road Show may be a critical open- 
ing wedge, but not an end in itself. 

e. A team of men will be assigned to a potential client until 
the prospect is realized or put aside. 

f. Individual sales strategies must be approved and inte- 
grated into over-all plans. 

4. What will happen to maintain enthusiasm or interest after 
the Road Show? 

a. Periodic mailings of literature of all types. 

b. Invitations to be present at programs, where our speakers 
are appearing. 

c. Solicit invitations to make examinations and proposals. 

d. Cooperative ventures in areas that can produce business. 

e. Consistent effort to activate potential customers in some 
program of ours. 

f. Frequent social, personal, and business contacts calcu- 
lated to build a bridge of confidence in our ability, and 
find out where the jobs are and who can assist us in get- 
ting them. 

g. Availability to the potential customer, sensitivity to his 
needs, and awareness of when is the right time to press 
for a decision. 

C. Road Show 

1. What is the fundamental purpose of the Road Show? 

a. To be a part of a general sales strategy leading step by 
step to a situation where we can make a proposal. 

b. Tell what we do; create confidence in our ability to do 
it. 



Davis Engineering Company 263 

c. Create interest, good will, and meet customers face to 
face. 

d. Strengthen our sales approach by activating staff. 

e. Provide an antidote for machinery manufacturers' opin- 
ions about our role in industry. 

2. What is the budget? 

a. Each out of town show will require two to five days (in- 
cluding travel). 

b. Only a portion of the written or visual material will be 
reusable. 

c. The average Road Show will cost 

(1) Time preparing $1,000.00 

(2) Time away 1,200.00 

(3) Sales aids and amortization 400.00 

(4) Rent, food, etc., (25 people) 650.00 

$3,250.00 

d. Twenty shows per year per sales-technical team is a con- 
servative estimate of annual capability, if more personnel 
are added to the sales organization. 

e. We need about 22 new jobs in 1957 to support the pro- 
posed annual budget. 

3. Who will put on the Road Show? 

a. Arrangements will be a complex problem. 

b. Sales-technical staff should be active show participants. 

c. Minimum of two men are required to present a Road 
Show; if the job potential is good, or if the group contains 
very high level personnel, a principal speaker should be 
provided. 

d. Personnel assigned to Road Shows must be temporarily 
or semi-permanently disassociated from other responsi- 
bilities. 

e. Members of Road Shows must have wide experience and 
present mature, sound appearance to potential customers. 

4. What is the suggested theme for the Road Shows? 

a. Must be tied into literature, visual aids, and so on. 

5. Are we doing the Road Show in preference to some other, 
perhaps more effective tool? 

a. Road Shows should not replace or eliminate any existing 
effort. 

b. Road Shows should be one, new type of effort, the value 
of which can only be judged by results. 

c. The conception of the Road Show must be flexible and 
change as circumstances and experience require. 

6. What other things might we consider now or in the future? 



264 Davis Engineering Company 

a. Educational course, seminars, workshops 

(1) On Davis premises 

(2) On customer premises 

b. Cooperative efforts to find applications or solutions to 
industry problems 

c. Advertising program 

d. Mailing list and letter campaign 

e. Personal contact campaign 

IV. General Conclusions and Definition 

A Road Show must be a part of an enlarged sales effort, each show- 
ing specifically tailored for the potential customer. 
Contact with prospective customers must be established prior to a 
Road Show. 

The Road Show should be carried only to contacts where we are 
reasonably sure of having an audience we want. 
Now is the time for a Road Show because we have a large repertoire 
of impressive jobs. 

We define a Road Show as a tour of sales-technical personnel who 
meet the potential customer face to face, and use a pre -determined 
format to present our sales story. 

EXHIBIT TWO 
ABSTRACT OF CONSULTANT REPORT 

I. What steps were taken to conduct this study? 
II. What came out of investigations? 

III. Conclusions with reference to assignment. 

IV. What are the alternatives? 

A. Pursue the Road Show idea with support up to maximum 
budget will allow. 

B. Pursue a Road Show on a national scale with exploitation of 
existing contacts. 

C. Defer the Road Show but tailor promotional presentations for 
existing customer contacts. 

D. Defer the Road Show but step-up advertising and personal con- 
tact. 

V. Recommendation 

What Steps Were Taken to Conduct this Study? 
I. Personal Contacts. 

II. Conferences. 

A. Develop a sales strategy. 

B. Prepare recommendations. 



Davis Engineering Company 265 

What Came out of the Investigation? 

I. Sales strategy must grow out of experience and capabilities. 

II. Sales strategy must grow out of the potential market lor these 
services. 

III. Sales strategy must consider the necessity to be competitive. 

IV. The method of operation must support the sales strategy. 

V. Sales strategy must encompass a continuing and integrated effort. 
VI. The sales strategy must be compatible with company objectives. 

What Is Davis' Training and Experience? 

I. Training 

A. Command of professional tools 

B. Superior understanding of tooling problems 

II. Experience 

A. Large scale tooling problems 

B. Wide range of industries 

What Is the Potential Market? 

I. Firms Davis is currently working for or has worked for in the 

past. 

A. Competition has done 74 per cent of volume on repeat busi- 
ness. 

B. Davis has done most of its volume on repeat business. 

II. Other contacts already established 

A. No door opener needed. 

B. Existing expression of potential customer problems or interests. 

C. Some degree of rapport in existence. 

III. All other firms in need of our services 

A. No estimate of regional or national markets. 

B. No estimate of demand for some specialized tooling service. 

C. What is the desirable sales goal in terms of current budget? 
Development of annual volume for 1957 of $7,000,000 — need 
about 22 new jobs to make this billing. 

How Can Davis Become Competitive? 

I. Who is the competition? 

II. What are the essential characteristics of competitors' operations? 
A. Sales 

1. Social persuasion. 

2. Full time, high level salesmen. 



266 Davis Engineering Company 

3. Organized continuing promotional campaign. 

4. Unrelenting stimulation of customers. 

5. Build professional reputation of individual members of staff. 

B. Services 

1. Activities organized by regions which permit regular and 
satisfying communication with customers. 

2. Full time, permanent, well-equipped sales staff. 

3. Well-established method of project management. 

4. Draw on a wide variety and a long history of successful proj- 
ects. 

C. Cost of services 

1. Initial answers generated quickly. 

2. Out of pocket expenses minimized by regional organization. 

How Can the Method of Operation Support the Sales Strategy? 

I. Every staff member is a salesman. 

A. Every staff member must understand the sales strategy. 

B. Every staff member must understand his role in sales. 

C. Every staff member must understand how to conduct himself 
with the customer. 

II. Project work must satisfy the customer — Davis must do what it has 
agreed to do. 

A. Davis must do its work within the quoted time schedule. 

B. Each staff member must have an individual schedule which he 
is responsible for meeting. 

C. Content of a program must be compatible with what Davis has 
agreed to do. 

D. Staff members must be assigned only to jobs for which they have 
demonstrated capabilities. 

E. Davis must have records of what transpired on the job. 

III. Project work must be concluded so customer can make continued 
profitable use of new tools. 

A. Customer's personnel must understand what has been done. 

B. Customer's organization must be prepared to handle new tools. 

Sales Strategy Must Encompass a Continuing And Integrated Effort 

I. Permanent full-time administration of sales — no lapse in personal 
contacts. 

A. A Steady stream of new customer contacts and proposals. 

B. Constant personal follow-up. 

II. Regularly scheduled promotional activities of varied types. 
A. Distribution of literature. 
1. Quarterly bulletin. 



Davis Engineering Company 267 

2. Annual Report. 

3. Bi-monthly list of accomplishments. 

4. Periodic special mailings. 
B. Personal presentations. 

1. Road Shows. 

2. Professional societies and trade groups. 

III. Provide for continuous revision of sales tools. 

A. Measure effectiveness of sales tools — change and modify em- 
phasis of sales program. 

B. Add new tools as they are developed. 

IV. There must be an underlying and general plan. 

A. Objective: to get a steady stream of new contracts. 

B. Central themes must have wide applicability to the potential 
market. 

C. All parts of the plan must be compatible with central themes. 

D. Take advantage of accepted and proven methods of reaching 
potential customers. 

V. There must be basic selling techniques for handling contacts. 

A. Objective: to develop a tailored sales plan for each contact. 

B. There must be a scheduled step-by-step plan leading to a sale — 
there must be continuous revisions in light of each contact. 

C. All recognized sales techniques should be used where applica- 
ble: 

1. Social and business visits 

2. Presentations and Road Shows 

3. Demonstrations and workshops 

4. Cooperative efforts 

5. Promotional devices 

6. Distribution of literature 

Conclusions 

I. The Road Show has a specific place in an integrated sales program, 
but has little value by itself. 

II. Sales will be registered only by a well-planned, perservering pro- 
gram. 

III. Ability to close sales rests primarily on confidence in the company 
generated by extensive personal contact. 

IV. The market for Davis' services, in terms of past and potential 
contacts, indicates that the business should operate on a national 
scale. 



268 Davis Engineering Company 

QUESTIONS 

1. What is the system under study? 

2. What are the assumptions implicit in Davis's position: 

3. If you were Davis, what would you do? 

4. What are the systems aspects of the problem? 



Refe 



rences 



OPERATIONS RESEARCH 

Batchelor, James H., Operations Research: A Preliminary Annotated 
Bibliography. Cleveland: The Case Institute of Technology, 1951. 

Case Institute Proceedings in Operations Research. 

Charnes, A., W. W. Cooper, and A. Henderson, An Introduction to 
Linear Programming. New York: John Wiley & Sons, 1955. 

Churchman, C. West, Russell L. Ackoff and E. Leonard Arnoff, Intro- 
duction to Operations Research. New York: John Wiley & Sons, 
1957. 

Ireson, W. Grant, and Eugene L. Grant, Handbook of Industrial En- 
gineering and Management. Englewood Cliffs, N.J.: Prentice-Hall, 
Inc., 1955. 

Journal of the American Institute of Industrial Engineers. 

Journal of the Operations Research Society of America. 

McClosky, J. F., and F. N. Trefethen, Operations Research for Manage- 
ment, two vols. Baltimore: Johns Hopkins Press, 1954. 

Operational Research Qiiarterly. 

Publications of the American Management Association. 

Publications of the Institute of Management Sciences. 

Publications of the Management Sciences Research Project, University 
of California at Los Angeles. 

Publications of the Office of Naval Research, Washington, D.C. 

Publications of the Rand Corporation. 

Vazsonyi, A., Scientific Programming in Business and Industry. New 
York: John Wiley & Sons, 1958. 

ELECTRONIC DATA PROCESSING 

Canning, R. G., Cutting the Cost of Your EDP Installation. Los Angeles: 
Published by the author, 1958. 

269 



270 References 

Canning, R. G., Electronic Data Processing for Business and Industry. 
New York: John Wiley & Sons, 1956. 

Canning, R. G., Installing Electronic Data Processing Systems. New 
York: John Wiley Sc Sons, 1957. 

Canning, Sisson, and Associates, "Suggested Reading List," Data Process- 
ing Digest, February, 1957. 

Handbook of Automation, Computation and Control, three vols. New 
York: John Wiley Sc Sons, 1959. 

Kozmetsky, G., and P. Kircher, Electronic Computers and Management 
Control. New York: McGraw-Hill Book Company, 1956. 

Publications of the Association for Computing Machinery. 

Publications of the Institute of Radio Engineers. 

RELATED TOPICS 

Bross, I. D. F., Design for Decision. New York: The Macmillan Com- 
pany, 1953. 

Chicago Area Transportation Study, Survey Findings, Vol. I, State of 
Illinois, December, 1959. 

Fisher, R. A., The Design of Experiments. New York: Hafner Publish- 
ing Co., 1951. 

Goode, H. H., and R. E. Machol, System Engineering. New York: Mc- 
Graw-Hill Book Company, 1957. 

Payne, S., The Art of Asking Questions. Princeton, N.J.: Princeton 
University Press, 1951. 

Publications of the Society for General Systems Research. 

Wiener, Norbert, Cybernetics. New York: John Wiley R: Sons, 1948. 






Ind 



ex 



Abstractions, in business systems, 60-63 

Access time, 115 

Accumulator, 115 

Accuracy, in system design, 14 

Action Notices, to supplement feedback, 

65 
Adder, 115 
Adding machines, 93 
Address, term, 115 
Address part of instruction, 115 
Address system, 115, 138 
Administration, as nonphysical system, 7 
Alpha data, 120 

Analysis of Scope of Study (chart), 78 
Ashby, W. Ross, 4 

Assignment, in system studies, 71-74 
Automatic operation, of electronic data 

processing, 98-99 



Ballistic missiles see Missiles 

Beaver Alliance Aircraft Corp., case study, 

227-237 
Bell electronic data processors, 95 
Benchmark problems, in system design, 

45 
Binary number system, 115, 120, 133, 138 
Bit, term, 115 

"Black box" concept, 3-5, 17-18 
Boards of directors, as system purchasers, 

53 
Boundaries, concept of, 20-26, 30-31, 38, 

49, 55, 74-77 
Branching, term, 115 
Budgets, as feedback, 17 
Buffers, 98, 115, 139 
Burroughs equipment, 127 
Business analysis, in feasibility studies, 

112-113 



Business machines, 93-94 
Business systems, investigations of, 



1,41 



Caratron devices, 123, 124 

Carlysle, Inc., case study, 209-217 

Cash registers, 93 

Central data processing facility, 48, 101— 

105 
Character-at-a-Time Printer, 124 
Checklist, for setting up assignments, 82- 

86 
Climate, as filter, 27 
Closed shop method, of using electronic 

data processing equipment, 103 
Closed systems, hypothesis of, 3 
Combined Release Schedule (chart), 186 
Completeness of data, in system design, 

14 
Compute orthocount, 128 
Computers see Electronic computers and 

data processing 
Conceptual models see Models, con- 
ceptual 
Console typewriter, 136 
Controls, 11-13, 16, 23, 28, 60, 61, 63, 161 

in electronic data processing, 99 

feedback as, 60-61 

mechanisms, 66-70 

systems analysts as, 40 

system purchasers as, 52-54 

term, 115 
Correlation and regression analysis, 159- 

160 
Cost-plus-fixed fee contracts, 75 
Costs: 

of data processing, 42, 43 

malfunctions, 21 

one-time vs recurring, 146-148 

reduction study, 222-226 



271 



272 Index 



Costs (cont.) 

savings, 142-146 

of systems, 142-155 
Customer Schedule (chart), 185 



DEW, 145 
Data: 

Collection Form, 108 

costs, processing, 42, 43, 98 

deterioration, 4 

in information theory, 59 

processing system design, in feasibility 
studies, 113-114 

processors, electronic see Electronic 
computers and data processing 

specialists, processing, 44 

Specification Sheet, 61 
DATAmatic 1000 computer, 126 
Datatron 220 computer, 119-125 

characteristics, 119-120 

input methods, 123-124 

instruction system, 121-122 

number system, 120 

output methods, 124-125 

storage, 122-123 
Davis Engineering Co., case study, 256- 

268 
Decision making, in electronic data proc- 
essing equipment, 100-101 
Descriptions of systems, 20-40 
Descriptive statistics, in operations re- 
search, 158 
Design of systems see Systems design 
Detail Computer Flow Charts, 88, 89 
Detail Machine Operation Flow Charts, 

88 
Detail Process Flow Charts, 87 
Diagramming techniques, 28-30 
Digital storage device, 96 
Distributed read-write, 128 
Drawing Room Manuals, 102 



EDSAC, 96 

EDVAC, 96 

ENIAC, 95, 197 

ERMA, 100 

Eckert, 95 

Edit-input and output, as filters, 27, 61 
^Effectiveness, of system design, 55-59 

Electronic computers and data process- 
ing, 19, 30, 44-45, 47-48, 68-69, 93- 
117, 160, 197-198, 240-241 



Electronic computers and data process- 
ing (cont.) 

automatic operation, 98-99 

capabilities, 48, 96 

decision making, 100-101 

development of, 95-96 

early processors, 93-95 

evaluation of, 118-141 

flexibility, 99-100 

glossary of terms, 115-117 

limitations, 97 

organization of, 9, 11 

organizing to centralize, 101-105 

preliminary surveys, 109-112 

as processors, 47-48 

requirement for, 10 

scheduling for, 48 

sorting, 97 

speed, 97-98 

as systems, 9-10 
Elements, of systems, 8-9, 14 
Engineering, as nonphysical system, 7 
English language programming, 99 
Environment, as filter, 26-27 
Environment, as processor, 25 
Equipment evaluation, in feasibility 

studies, 114-115 
Equipment manufacturers, 106 
Equipment systems, evaluation of, 118— 

141 
Errors, 66-70, 94, 95 
Existing methods, investigation of, 31-32 



Factor analysis, 160-161 

Factories, 12-18 

Feasibility, term, 105 

Feasibility studies, 36, 46-47, 105-109, 

112-115 
Feedback. 11, 13-14, 16-17, 23. 28, 35-36, 
65-66 

as control, 60-61 

filtering of, 26-28 

loops, design of, 63-66 

term, 116 
Files, of input transactions, 30 
Filter concept, defined, 27 
Filtering, of input and feedback, 26-28, 

61 
Finished goods inventories, 57-58 
Fixed word length, 117 
Fleet, A. B., and Co., case study, 238-245 
Flexibility, of electronic data processing 

systems, 99-100 
Flexowriter, 130, 131 



Index 273 



-^Flow charts, 28, 34, 41, 70, 86-90 

Detail Computer, 88, 89 

Detail Machine Operation, 88 

Detail Process, 87 

Master Flow Diagram, 80, 81, 86-87 

symbols for, 88 

Top Computer, 88 

Top Process Flow Chart, 87, 88 
Forecasting, 250-255 
Frequency, as system characteristic, 14 



General systems theory, and operations 

research, 165-166 
Geographical location, as filter, 27 



Headcounts, 80-81 
High Speed Printer, 140 
High Speed Reader, 139, 140 
Honeywell 800 computer, 125-132 

characteristics, 125-126 

input methods, 130-131 

instruction system, 127-129 

number system, 126-127 

output methods, 131-132 

storage, 129-130 
Human factors, adjustments to, 40 
Fluman factors, as variables, 6 
Hypotheses, in system design, 35-37 



Institutions, 25 

Instruction, term, 116 

Instruction codes, for electronic data 

processing, 99 
Integrated data processing system for 

Production Control (chart), 62 
Integrated data processing system using 

punched cards or computer as the 

processor (chart), 64 
Integration, of systems, 17-19, 29-30, 48, 

59 
Internal review checklists, 84-86 
International Corp., case study, 204-208 
Interviewing, for data collection, 32, 33- 

34 
Inventory files, and feedback, 66-68 
Inventory status, as criteria for schedul- 
ing system, 56-58 
Inventory status, as input, 16 
Investigations, of systems, 31-35, 41 
Isomorphic systems, 25-26 



LOGBALNET, military problem, 105 

Labor, as input, 16 

Language programming, 99 

Lee Co., case study, 180-192 

Linear programming, 160 

Listing machines, 93 

Logical components, term, 116 

Los Angeles, Calif., 27 



IBM 305 Ramac, 132-136 

characteristics, 132-133 

input methods, 135-136 

instruction system, 133-134 

number system 133 

output methods, 136 

storage, 134-135 
IBM 323 machine, 136 
IBM 370 Printer, 136 
IBM 382 Reader, 136 
IBM 1401 computer, 137 
IBM equipment, 105, 126, 127, 140 
IBM printers, 136 
IBM punched card machines, 94, 123, 124, 

131 
IBM tape-to-card converter, 131 
Implementation, as step in system design, 

38-40 
Incompletely structured systems, 6-9 
Information theory, 59 
Inputs, 11, 12,22,28,47 

filtering of, 26-28 

term, 116 



Machine load chart, 189 
Machine verification, 61 
Magnetic Character Readers, 131 
Magnetic tape, 47, 98, 125, 129-130 
Malfunctions in systems, 21 
Man-machine systems, 6, 10, 45, 94 
Man-made objects, 25 
Management, 41, 76, 79, 92, 93, 101-105, 

144, 145, 146, 147 
Manual systems, 58-59 
Manufacturing — Calender Day Cross 

Reference Chart, 190 
Manufacturing Industries, W. S. G., 209 
Mark III calculator, 95 
Mark IV calculator, 95 
Marxson and Co., case study of method 

of operation, 80-82, 193-203 
Master Flow Diagram, 80, 81. 86-87 
Mathematical Laboratory, University of 

Cambridge, 96 
Mathematical models, 36, 157-158 
Mauchley, 95 



274 Index 



Mechanization, in system design, 45-47 

Memory, term, 116 

Missiles, 10-12, 53, 55-56 

Models, conceptual, 6, 34-37, 60, 61, 69- 

70 
Modification, term, 116 
Module, term, 11 
Morse, Dr. Philip, 157 
Multi-Program Control Section, 130 
Multiple regression analysis, 159-160 
Multiple transfer, 128 

National Cash Register computers, 141 
Nonphysical systems, 6-8 

"One-run approach," in solid-state com- 
puters, 137 
Open shop method, of using electronic 

data processing equipment, 102 
Operation codes, 116 
Operations research, 19, 48, 156-166 

background, 156-158 

control system analysis, 161 

correlation and regression analysis, 
159-160 

descriptive statistics, 158 

factor analysis, 160-161 

function and methodology, 157 

and general systems theory, 165-166 

linear programming, 160 

mathematics, 157-158 

simulation, 161-162 

statistical sampling and inference, 158- 
159 

and systems analysis, 162-165 

techniques of, 158-162 
Optimal system design, 39, 50-51, 70 
Order release subsystems, 15 
Organization, of systems, 9-12 
Organization charts, 79 
Orthotronic control, 129, 131 
Outlines, indentures for, 34 
Outputs, 11, 12,28,47 

effectiveness of, 55-56 

requirements, 30 

term, 116 

testing of, 55-56 



Physical systems, 3-6 
Physical systems-business systems anal- 
ogy, 19 
Pilot operations, 38-40 
Policing period, after implementation of 

system, 40 
Postulated systems, 36, 56, 60-70, 107- 

109, 114-115 
Preparations for systems studies, 71-92 
assignments, 71-74 
defining problem area and boundaries. 

74-77 
method of operation, 80-82 
outlines, 78-79 

priority of effort and schedule, 77-79 
selling the assignments, 92 
staffing of project, 90-92 
team work, 90-92 
Prime contractors, 51 
Printed reports, 132 
Printer-Punch, 132 

Processors (See also Electronic comput- 
ers and data processing), 11, 12. 15- 
16, 20-22, 25, 28, 42, 44, 47, 60, 63 
computers as, 47-48 
defined, 25 

electronic data, 95-96 
term, 116 
Product users, as system purchasers, 54 
Production control, 15-16, 20, 21, 22, 25, 

58, 75, 76, 77 
Production control subsvstems. inputs in. 

17 
Production Requirements Report (chart), 

187 
Production system (chart), 182 
Production Tickets, 230-233 
Profit, as output, 51 
Programming, 98-99, 116 
Progress reports, as feedback, 16 
Properties, of systems, 8-9 
Punched cards, 46. 47, 48, 68, 94. 131-132. 

135, 136, 139 
Punched paper tape, 94-95, 124, 130-131 
Punched Paper Tape Reader (IBM 382). 

136 
Purchasers of systems see Systems pur- 
chasers 



Paper tape, punched, 94-95, 130-131 
Paper Tape Punch, 124 
Parkinson's Law, 18 
Parts production routing (chart), 184 
Performance reports, as feedback, 17 
Photoreaders, 123 



Quality Control Manuals, 102 

Ramac see IBM 305 Ramac 
RCA computers, 136-137 
Raw materials inventory, 56-57 
Read, term, 116 



Index 275 



Read-Punch Unit, 140 
Record, term, 116 
Record storage unit, 116 
Redesigning, 38, 43-45, 49-50, 106 
Redundance, in information theory, 59 
Register, term, 116 
Regression analysis, 159-160 
Reliability, of system elements, 14 
Remington Rand computers, 140-141 
Reorder reaction, as feedback, 13 



Sales data, as feedback, 65-66 
Sampling techniques, and input, 14 
Savings, 142-146 

Schedule maintenance, 15-16, 54 
Scheduling, for computer systems, 48, 79 
Scrap reduction, as example of produc- 
tion control, 22-24 
Service Bureau Corporation, 133 
Shift and select instructions, 128 
Side effects, in system design, 48-50 
Simpson-East Corp., case study, 218-226 
Simulation, 161-162, 240-241 
Simulator instructions, 128-129 
Solid-state computer see UNIVAC Solid 

State computer 
Sorting, by computers, 97 
Specialists, in data processing, 44 
Speed, of electronic data processing, 97- 

98 
Spoken word, in data collection, 32 
Standard data processing symbols 

(chart), 88 
Static, in information theory, 59 
Statistical sampling and inference, 158— 

159 
Statistical validation, of systems, 37 
Steps Required to Implement a Com- 
puter Application After Postulated 
System Design (chart), 91 
Stockholders, as system purchasers, 53-54 
Storage, term, 116 
Storage location, 117 
Stored program machine, 117 
Structured systems, 3-6 
Subcontractors, requirements for, 51, 52 
Subsystems: 

boundaries of, 22 

compatibility, 37 

interdependence of, 80 

results of integration, 17 

systems purchasers for, 54 

term, 5 

testing of, 38 



Supervisory Printer unit, 124 

Surveys, of data processing see Feasibility 

studies 
Surveys, preliminary, for electronic sys- 
tems, 109-112 
Symbols, use of, 60 
Symbols for flow charts, 88 
Systems {Also passim) 
"—analysis, 18-19, 24 

analysis, relation to operations re- 
^^ search, 162-165 

analysts, as c ontrol elements, 40 

analysts, as processors, 54 

approach, 40 

design see Systems design infra 

diagrams, 28-30 

illustrations of, 28-30 

integration see Integration, of systems 

modules, as symbols of qualitative 
ideas, 60 

purchasers, 47^18, 51-54, 55, 59, 77 

requirements, 20-21, 44, 45 

review checklists. 82-84 

studies, preparations for see Prepara- 
tions for system studies 
Systems design: 

alternatives, 42-48 

benchmark approach, 45 

computers, 44-45 

criteria, 14 

effectiveness, 55-59 

and feasibility studies, 106 

fundamentals, 41-59 

hypotheses, 35-37 

implementation, 38-40 

investigation, 31-35 

mechanization, 45-47 

one-for-one changeovers, 42-43 

optimal, 39, 50-51, 70 

philosophies of, 41 

redesigning, 43-45 

as result of Management policy, 41 

side effects, 48-50 

steps in, 30-40 



Tabulating equipment, 46 

Technical orders, and system purchasers, 

53 
Templar, G. W„ and Co., case study, 75- 

77, 80, 169-179 
Time, functions of, 21, 71-73, 71 
Timekeeping procedure, proposed, 233- 

236 
Top Computer Flow Charts, 88 



276 Index 



Top Process Flow Chart, 87, 88 
Traffic Control Section, 130 
Transistors, use of, 136-140 
Tucson, Arizona, 27 

UNISERVOS, 139 

UNIVAC magnetic tapes, 139 

UNIVAC Scientific, 141 

UNIVAC Solid-State computer, 136-140 

characteristics, 136-138 

input methods, 139-140 

instruction system, 138 

number system, 138 

output methods, 140 

storage, 138-139 
Unstructured systems, variance in, 37 

Variable word length, 117 



Variance, in unstructured systems, 37 
Von Neumann, 95 



Weapons systems, 5, 10-12, 50, 51, 53, 55- 

56 
Wesley Engineering, Inc., case study, 80, 

246-255 
Word, term in electronic data processing. 

117 
Word length, 117 
Work Assignment Form, 71-73, 79, 82, 

147 
Work-in-process inventories, 57 
World War II, and operations research. 

156-157 
Write, term in electronic data processing. 

117 
Written materials, in data collection, 32 



Date Due 
Due Returned Due Returned 




l> a 6 z 














65S.5| 

IfNGINP 
t Di- 
ll r r